Description
Problems
Please access the CSMAR database and download the following data for all listed firms in the A-share
market: (i) monthly Stock Closing Price, Return (without cash dividend reinvested) over Dec. 2009 to Dec.
2023 from Individual Stock Trading table, (ii) quarterly Return on Equity – TTM and Net Assets per Share
over 2009Q3 to 2023Q4 from Financial Indicator table, (iii) daily stock volatility (of the log return of the
lastest 250 trading days) at 2010/12/31 from Stock Market Derivative Index table. Then derive monthly P/B
ratios for all A-share stocks from Jan. 2010 to Dec. 2023. Exclude records with P/B ratios less than P/B
ratio’s 5th percentile or greater than P/B ratio’s 95th percentile.
1. Focusing on the observations for all A-share firms at the end of 2010, regress the P/B ratio at Dec.
2010 on Return on Equity – TTM (ROE) at 2010Q4 and Stock Volatility at 2010/12/31, i.e., you estimate
the following cross-sectional regression:
P/Bi = α + β1ROEi + β2Stock Volatilityi + ϵi
.
Report regression results and discuss your findings.
2. For each month from Jan. 2010 to Dec. 2023, sort firms based on their last-month P/B ratios and
divide firms into ten groups according to last-month P/B ratio deciles. By holding all stocks within
each group with equal weights and rebalancing positions every month, we can construct ten portfolios.
Please calculate the monthly returns for the ten portfolios and use a bar chart to illustrate average
returns for the ten portfolios from Jan. 2010 to Dec. 2023, and discuss your findings.
Hints
1. A cross-sectional data set is consisted of observations for different listed companies at a single time
point.
2. Note that financial statements are usually reported quarterly while stocks are traded every trading day.
To construct monthly valuation measures, you may divide the closing price with the latest accounting
indicator. For example, you may construct the P/B ratio for company i at 2019m11 (i.e., Nov. 2019)
as follows:
P/B i,2019m11 =
Closing pricei,2019m11
Net Assets per Sharei,2019q3
.
3. You may exclude parent statements from financial indicator data.
4. In Problem 2, you are expected to construct ten portfolios based on P/B ratiosin the previous month. In
other words, the composition of these portfolios changes every month. Denote by Di,t(i ∈ {1, 2, · · · , 9})
the 9 deciles/cutoffs for P/B ratios at t and further define D0,t (D10,t) as the minimal (maximal)
P/B ratios at t.
Then portfolio i at t consists of equal-weighted stock j’s with P/B ratioj,t−1 ∈
[Di−1,t−1, Di,t−1] and the return for the i-th portfolio at month t (denoted by r
p
i,t
) is given by
r
p
i,t =
1
Ni,t
Ni,t
∑
j=1
r
s
j,t
,
where Ni,t denotes the number of stocks with P/B ratios at t − 1 lying in [Di−1,t−1, Di,t−1] and r
s
j
denotes the monthly return for stockj at t.
You may find bysort and xtile in Stata helpful to generate P/B ratio deciles by month.
