Replicating Ang, Shtauber & Tetlock (2013): Asset Pricing in the Dark¶

U.S.E. Finance Data Hub / WRDS Database: OTC Markets¶

Paper replicated: Ang, A., Shtauber, A. A. & Tetlock, P. C. (2013). Asset Pricing in the Dark: The Cross-Section of OTC Stocks. The Review of Financial Studies, 26(12), 2985–3028.

Database used: OTC Markets (Data Hub guide)

---

What this notebook does¶

Ang, Shtauber & Tetlock (2013) study asset pricing in over-the-counter (OTC) equity markets, which are less liquid and less transparent than listed exchanges. They find that standard asset pricing factors (market, size, value) partially explain OTC stock returns, but that illiquidity and information asymmetry play a much larger role than on listed exchanges.

The WRDS OTC Markets database provides security-level data for OTC-traded stocks.

1. Pull OTC security data from OTC Markets (otc) via the WRDS API
2. Explore the OTC market structure (OTCQX, OTCQB, Pink tiers)
3. Examine the distribution of OTC-traded securities by market tier and trading activity
4. Compare the characteristics of OTC stocks to the paper's findings

Learning objectives¶

• Practice connecting to WRDS and querying the OTC Markets database
• Understand the structure of U.S. OTC equity markets and their tiered system
• Learn how OTC markets differ from listed exchanges in terms of liquidity and transparency
• Build intuition for why asset pricing in less-liquid markets requires different approaches

Requirements to run this notebook¶

• A valid WRDS account with OTC Markets access (ask the Finance Data Hub if you don't have one yet)
• pip install wrds pandas numpy matplotlib seaborn
• You will be prompted for your WRDS username/password the first time you connect (or set up a .pgpass file, see the WRDS Python guide)

1. Setup and WRDS connection¶

In [ ]:

import wrds
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

sns.set_theme(style="whitegrid", palette="deep")
pd.set_option("display.max_columns", 50)

db = wrds.Connection()

import wrds import pandas as pd import numpy as np import matplotlib.pyplot as plt import seaborn as sns sns.set_theme(style="whitegrid", palette="deep") pd.set_option("display.max_columns", 50) db = wrds.Connection()

2. Exploring the OTC Markets database¶

The OTC Markets data is accessed through the otc library. Since the exact table names may vary, we first discover the available tables and their structure.

OTC Markets Group classifies securities into three tiers based on disclosure and quality standards: OTCQX (highest quality, meeting financial standards), OTCQB (venture market, current SEC reporting), and Pink (speculative, varying levels of disclosure).

In [ ]:

# Discover the available tables
tables = db.list_tables(library="otc")
print("Available tables in otc library:")
print(tables)

# Discover the available tables tables = db.list_tables(library="otc") print("Available tables in otc library:") print(tables)

In [ ]:

# The only table is otc.endofday. Let's inspect its columns.
desc = db.describe_table(library="otc", table="endofday")
print(desc[["name"]].to_string())

# The only table is otc.endofday. Let's inspect its columns. desc = db.describe_table(library="otc", table="endofday") print(desc[["name"]].to_string())

3. Cleaning the sample¶

In [ ]:

# The date column is closingbestbiddate (dad_pal is mostly null).
# Pull data using closingbestbiddate as the date filter.

query = """
    SELECT secid, symbol, cusip, sectype, tiername, caveatemptor,
           closingbestbiddate,
           openprice, highprice, lowprice, lastprice,
           previouscloseprice, sharevolume, dollarvol, tradecount,
           closinginsidebidprice, closinginsideaskprice,
           insidebid_askmidprice, sharesoutstanding
    FROM otc.endofday
    WHERE closingbestbiddate >= '2018-01-01'
      AND closingbestbiddate <= '2018-12-31'
      AND lastprice IS NOT NULL
"""

df = db.raw_sql(query, date_cols=["closingbestbiddate"])
df = df.rename(columns={"closingbestbiddate": "date"})
print(f"Rows pulled: {len(df):,}")
print(f"Date range: {df['date'].min()} to {df['date'].max()}")
print(f"Unique securities: {df['symbol'].nunique():,}")
df.head()

# The date column is closingbestbiddate (dad_pal is mostly null). # Pull data using closingbestbiddate as the date filter. query = """ SELECT secid, symbol, cusip, sectype, tiername, caveatemptor, closingbestbiddate, openprice, highprice, lowprice, lastprice, previouscloseprice, sharevolume, dollarvol, tradecount, closinginsidebidprice, closinginsideaskprice, insidebid_askmidprice, sharesoutstanding FROM otc.endofday WHERE closingbestbiddate >= '2018-01-01' AND closingbestbiddate <= '2018-12-31' AND lastprice IS NOT NULL """ df = db.raw_sql(query, date_cols=["closingbestbiddate"]) df = df.rename(columns={"closingbestbiddate": "date"}) print(f"Rows pulled: {len(df):,}") print(f"Date range: {df['date'].min()} to {df['date'].max()}") print(f"Unique securities: {df['symbol'].nunique():,}") df.head()

4. Analyzing OTC market characteristics¶

Ang et al. (2013) document several key features of OTC stocks:

1. OTC stocks are significantly less liquid than exchange-listed stocks (wider bid-ask spreads, lower volume)
2. OTC stocks have higher return volatility
3. Standard asset pricing factors explain less of the cross-section of OTC returns
4. Illiquidity is a priced factor in OTC markets

We examine the first two features directly from the OTC data. The asset pricing tests would require merging OTC returns with Fama-French factors from the ff_all database.

In [ ]:

# Clean: require positive price and volume
df = df.dropna(subset=["lastprice", "sharevolume"])
df = df[(df["lastprice"] > 0) & (df["sharevolume"] > 0)]

print(f"Observations after cleaning: {len(df):,}")
print(f"Unique securities: {df['symbol'].nunique():,}")
print(f"Date range: {df['date'].min()} to {df['date'].max()}")

# Bid-ask spread (a key illiquidity measure in Ang et al.)
df["spread"] = df["closinginsideaskprice"] - df["closinginsidebidprice"]
mid = df["insidebid_askmidprice"].fillna(
    (df["closinginsidebidprice"] + df["closinginsideaskprice"]) / 2
)
df["spread_pct"] = df["spread"] / mid

print(f"\nPrice distribution:")
print(df["lastprice"].describe())
print(f"\nVolume distribution:")
print(df["sharevolume"].describe())

# Classify by price tier
df["price_tier"] = pd.cut(df["lastprice"],
                          bins=[0, 0.01, 0.10, 1, 5, 50, 1e6],
                          labels=["<$0.01", "$0.01-0.10", "$0.10-1", "$1-5", "$5-50", ">$50"])
print("\nSecurities by price tier:")
print(df.groupby("price_tier")["symbol"].nunique())

# Summary by OTC market tier
print("\nSecurities by OTC Market Tier:")
print(df.groupby("tiername")["symbol"].nunique().sort_values(ascending=False))

print("\nMedian Price, Volume, and Spread by Tier:")
tier_stats = df.groupby("tiername").agg(
    median_price=("lastprice", "median"),
    median_volume=("sharevolume", "median"),
    median_spread_pct=("spread_pct", "median"),
    n_securities=("symbol", "nunique")
).sort_values("n_securities", ascending=False)
print(tier_stats.round(4))

# Compute daily returns per security
df = df.sort_values(["symbol", "date"])
df["ret"] = df.groupby("symbol")["lastprice"].pct_change()
df = df[df["ret"].between(-0.99, 10)]

# Clean: require positive price and volume df = df.dropna(subset=["lastprice", "sharevolume"]) df = df[(df["lastprice"] > 0) & (df["sharevolume"] > 0)] print(f"Observations after cleaning: {len(df):,}") print(f"Unique securities: {df['symbol'].nunique():,}") print(f"Date range: {df['date'].min()} to {df['date'].max()}") # Bid-ask spread (a key illiquidity measure in Ang et al.) df["spread"] = df["closinginsideaskprice"] - df["closinginsidebidprice"] mid = df["insidebid_askmidprice"].fillna( (df["closinginsidebidprice"] + df["closinginsideaskprice"]) / 2 ) df["spread_pct"] = df["spread"] / mid print(f"\nPrice distribution:") print(df["lastprice"].describe()) print(f"\nVolume distribution:") print(df["sharevolume"].describe()) # Classify by price tier df["price_tier"] = pd.cut(df["lastprice"], bins=[0, 0.01, 0.10, 1, 5, 50, 1e6], labels=["<$0.01", "$0.01-0.10", "$0.10-1", "$1-5", "$5-50", ">$50"]) print("\nSecurities by price tier:") print(df.groupby("price_tier")["symbol"].nunique()) # Summary by OTC market tier print("\nSecurities by OTC Market Tier:") print(df.groupby("tiername")["symbol"].nunique().sort_values(ascending=False)) print("\nMedian Price, Volume, and Spread by Tier:") tier_stats = df.groupby("tiername").agg( median_price=("lastprice", "median"), median_volume=("sharevolume", "median"), median_spread_pct=("spread_pct", "median"), n_securities=("symbol", "nunique") ).sort_values("n_securities", ascending=False) print(tier_stats.round(4)) # Compute daily returns per security df = df.sort_values(["symbol", "date"]) df["ret"] = df.groupby("symbol")["lastprice"].pct_change() df = df[df["ret"].between(-0.99, 10)]

5. Visualizing the results¶

In [ ]:

fig, axes = plt.subplots(1, 2, figsize=(14, 5))

# Price distribution (log scale)
axes[0].hist(np.log10(df["lastprice"].clip(lower=0.001)), bins=50,
             color="steelblue", edgecolor="white")
axes[0].set_title("Distribution of OTC Stock Prices (log10 scale)\n(OTC Markets, 2018)")
axes[0].set_xlabel("log10(Price)")
axes[0].set_ylabel("Number of observations")

# Volume distribution (log scale)
axes[1].hist(np.log10(df["sharevolume"].clip(lower=1)), bins=50,
             color="darkorange", edgecolor="white")
axes[1].set_title("Distribution of OTC Daily Trading Volume (log10 scale)")
axes[1].set_xlabel("log10(Volume)")
axes[1].set_ylabel("Number of observations")

plt.tight_layout()
plt.show()

fig, axes = plt.subplots(1, 2, figsize=(14, 5)) # Price distribution (log scale) axes[0].hist(np.log10(df["lastprice"].clip(lower=0.001)), bins=50, color="steelblue", edgecolor="white") axes[0].set_title("Distribution of OTC Stock Prices (log10 scale)\n(OTC Markets, 2018)") axes[0].set_xlabel("log10(Price)") axes[0].set_ylabel("Number of observations") # Volume distribution (log scale) axes[1].hist(np.log10(df["sharevolume"].clip(lower=1)), bins=50, color="darkorange", edgecolor="white") axes[1].set_title("Distribution of OTC Daily Trading Volume (log10 scale)") axes[1].set_xlabel("log10(Volume)") axes[1].set_ylabel("Number of observations") plt.tight_layout() plt.show()

In [ ]:

# Compute per-security summary statistics
sec_stats = df.groupby("symbol").agg(
    median_price=("lastprice", "median"),
    median_volume=("sharevolume", "median"),
    volatility=("ret", "std"),
    n_days=("date", "nunique"),
    tier=("tiername", "first")
)
# Keep only securities with at least 50 trading days
sec_stats = sec_stats[sec_stats["n_days"] >= 50]
print(f"Securities with 50+ trading days: {len(sec_stats):,}")

fig, axes = plt.subplots(1, 2, figsize=(14, 5))

# Volatility distribution
axes[0].hist(sec_stats["volatility"].clip(upper=0.5) * 100, bins=50,
             color="firebrick", edgecolor="white")
axes[0].set_title("Distribution of Daily Return Volatility\n(OTC Securities, 2018)")
axes[0].set_xlabel("Daily Volatility (%)")
axes[0].set_ylabel("Number of securities")

# Volatility by OTC tier
tier_vol = sec_stats.groupby("tier")["volatility"].median() * 100
tier_vol = tier_vol.sort_values(ascending=False)
axes[1].barh(tier_vol.index, tier_vol.values, color="steelblue")
axes[1].set_title("Median Daily Volatility by OTC Market Tier")
axes[1].set_xlabel("Median Daily Volatility (%)")

plt.tight_layout()
plt.show()

# Summary: average volatility by price tier
tier_vol_price = df.groupby("price_tier")["ret"].std() * 100
print("\nAverage Daily Volatility by Price Tier (%):")
print(tier_vol_price.round(2))

# Compute per-security summary statistics sec_stats = df.groupby("symbol").agg( median_price=("lastprice", "median"), median_volume=("sharevolume", "median"), volatility=("ret", "std"), n_days=("date", "nunique"), tier=("tiername", "first") ) # Keep only securities with at least 50 trading days sec_stats = sec_stats[sec_stats["n_days"] >= 50] print(f"Securities with 50+ trading days: {len(sec_stats):,}") fig, axes = plt.subplots(1, 2, figsize=(14, 5)) # Volatility distribution axes[0].hist(sec_stats["volatility"].clip(upper=0.5) * 100, bins=50, color="firebrick", edgecolor="white") axes[0].set_title("Distribution of Daily Return Volatility\n(OTC Securities, 2018)") axes[0].set_xlabel("Daily Volatility (%)") axes[0].set_ylabel("Number of securities") # Volatility by OTC tier tier_vol = sec_stats.groupby("tier")["volatility"].median() * 100 tier_vol = tier_vol.sort_values(ascending=False) axes[1].barh(tier_vol.index, tier_vol.values, color="steelblue") axes[1].set_title("Median Daily Volatility by OTC Market Tier") axes[1].set_xlabel("Median Daily Volatility (%)") plt.tight_layout() plt.show() # Summary: average volatility by price tier tier_vol_price = df.groupby("price_tier")["ret"].std() * 100 print("\nAverage Daily Volatility by Price Tier (%):") print(tier_vol_price.round(2))

6. Comparing to Ang, Shtauber & Tetlock (2013): What's the same, what's different¶

What replicates cleanly

The tiered structure of OTC markets (OTCQX, OTCQB, Pink) maps directly onto the database's classification system. The descriptive statistics on trading activity, price levels, and bid-ask spreads by tier should confirm the paper's finding that OTC stocks are dramatically less liquid than exchange-listed securities. The high concentration of very low-priced securities ("penny stocks") in the Pink tier is a distinctive feature of OTC markets.

Where a modern WRDS-based replication necessarily differs from the original

1. Return data. Ang et al. construct returns from OTC price data and apply careful filters for stale prices and data errors. The exact table and column names on WRDS may differ from those in the paper's data appendix. Students should use list_tables() and describe_table() to identify the correct fields.

2. Sample period. The paper covers 1977 to 2008, a period that includes both the old NASD OTC market and the modern OTC Markets Group structure. Our data reflects the current tiered system, which was formally introduced by OTC Markets Group in 2010.

3. Cross-listing. Many OTC stocks are foreign companies with primary listings abroad (ADRs). The paper carefully separates domestic OTC stocks from foreign cross-listings. Students should filter on security type to focus on domestic common stocks.

4. Asset pricing tests. The paper's main contribution is running Fama-MacBeth cross-sectional regressions with illiquidity as a factor. This requires combining OTC return data with the Fama-French factors from ff_all, which is a natural extension of this descriptive analysis.

---

References¶

• Ang, A., Shtauber, A. A. & Tetlock, P. C. (2013). Asset Pricing in the Dark: The Cross-Section of OTC Stocks. The Review of Financial Studies, 26(12), 2985–3028.
• WRDS OTC Markets guide: https://uufinance.github.io/data/wrds/databases/otc-markets/
• WRDS OTC Markets access page: https://wrds-www.wharton.upenn.edu/pages/get-data/otc-markets-group/
• WRDS Python/API setup guide: https://uufinance.github.io/data/wrds/notebook/

Prepared for the U.S.E. Finance Data Hub as a database-tutorial template.