TopAceEssays

In this assignment you are asked to replicate selected and slightly modified parts
of the article by Laurent Frésard titled “Financial Strength and Product Market
Behavior: The Real Effects of Corporate Cash Holdings”, Journal of Finance, Vol.
LXV, No. 3, 2010. The article is available at the following link:
http://onlinelibrary.wiley.com/doi/10.1111/j.1540-6261.2010.01562.x/abstract

1. Data preparation. Follow the steps described in section II.C of the article
   to construct a yearly panel of manufacturing firms that includes all
   necessary data items to run regression (1) on p. 1102 of the article. Use
   data for the sample period 1973-2007, i.e. one more year than the original
   paper.
   From the raw COMPUSTAT data, construct the dependent and
   independent variables needed to estimate equation (1). To construct the
   change in market share, do not follow the paper’s definition, but define
   the market share of firm i operating in industry j in a given year t as the
   fraction of firm i’s sales of the total sales in industry j in that year; the
   change in market share is then simply the year-to-year change of market
   share, market share in t minus market share in t-1. To construct the cash
   variable, simply divide cash holdings by assets. Do not subtract the
   industry average of cash holdings, and do not standardize cash by its
   industry-level standard deviation as it is done in the paper. The variable
   Tangibility is Property, Plant, and Equipment divided by total assets. (Do
   not use the paper’s definition of tangibility!). Check that your variables
   have reasonable summary statistics, and handle outliers where and as
   appropriate.
   In order to construct the import tariff variable, use the STATA file
   (“xm_sic87_72_105_20120424.dta”) which is available on Canvas with
   this assignment. It contains all relevant U.S. import data. For every year
   and four-digit SIC industry, calculate the percentage import tariff as the
   industry total of “duties charged” (variable name “duties”) divided by the
   industry total of dutiable import value (variable name “dutyval”). Define
   the year-on-year tariff change in an industry as the tariff in year t minus
   the tariff in year t-1 (not as the percentage change in tariffs, i.e. not
   dividing the difference by the past year’s tariff). Define the variable CUT2
   as a dummy variable that takes the value of one if import tariffs in a given
   industry fall by more than two times the median year-on-year tariff
   change in that industry. Equivalently, define a variable CUT25 (CUT3),
   taking the value of one if the decrease is more than 2.5 (3) times the
   2
   median change. Finally, merge the tariff data into the main data file.
2. Descriptive statistics. Provide a table (Table I) of descriptive statistics
   for the main variables used in the analysis: market share, year-on-year
   change in market share, cash holdings, tangibility, leverage, and firm size.
   For each variable, report the mean, median, standard deviation, the 5%
   and 95% percentile, and the number of observations. Briefly comment on
   the statistics in words.
3. Baseline panel regressions. Produce a table (Table II) with six columns
   showing panel regressions of the change in market share on the noninstrumented lagged cash holdings. Note that such a table is not shown in
   the original paper. In the first column, regress the change in market share
   on the once-lagged cash holdings only; in the second column, add the
   twice-lagged cash holdings as an independent variable; in the third
   column, add firm size, once- and twice-lagged leverage, once- and twicelagged market share growth; in column four, add year fixed effects; in
   column five, add firm fixed effects; in column six, run the same regression
   as in column five but replace year and firm fixed effects with industryyear fixed effects. Handle standard errors as you deem appropriate. In
   these regressions, do you find lagged cash holdings to be robustly
   correlated with market share? And does the magnitude of this
   relationship change with the controls being used?
4. IV regressions. Run IV regressions similar (but not identical) to those
   reported in Table 1 of the article. Use Stata’s ivreg2 or ivreghdfe
   commands (whichever is appropriate) to do this. For all IV regressions
   use the control variables and fixed effects that you used in column six of
   Table II. Produce a table (Table III) with two panels. Do not try to produce
   a table that looks similar to Table 1 in the paper. In the top panel (Panel
   A) report three second-stage regressions: the first one uses tangibility as
   the only instrument for cash holdings; the second one uses tangibility and
   once-lagged cash holdings as instruments; and the third one uses
   tangibility, one- and twice-lagged cash holdings as instruments for cash
   holdings. In the bottom panel (Panel B) report the three different first
   stage regressions that correspond to the second stage regressions of
   Panel A. (Your table will therefore look different than Table I of the
   article.) In a separate row at the bottom of that table, report the value of
   the F-statistic of the excluded instruments. What does this test tell you?
   And does the first stage pass that test? Do you find results similar to
   Frésard?
5. Difference-in-difference regressions. Run difference-in-difference
   regressions estimating equation (2) of the article. Somewhat differently
   from the original paper, report, in columns (1)-(3), difference-indifference regressions with the following independent variables only:
   Casht-1, CUTt, Casht-1× CUTt , as well as firm and industry-year fixed effects.
   Do not use twice-lagged cash as the original paper does. For column (1),
   use CUT2 as the treatment dummy; for column (2) CUT25; for column (3)
   3
   CUT3. What is the general interpretation of those three coefficients?
   Which is the coefficient of interest for this study? What is its
   interpretation? In columns (4)-(6), run the same regression as in (1)-(3),
   but add controls for firm size, once- and twice-lagged leverage, as well as
   once- and twice-lagged sales growth. Do you find results similar to
   Frésard? And do your findings from the IV and difference-in-difference
   regressions point in the same direction or not?
   Your answers to the above questions need to be in the following format:

• For part 1, you need to provide STATA code only. When you have
  completed the assignment, copy and paste the STATA code for the data
  preparation into your answer file.
• For each of questions 2-5, your answer should contain three pages: The
  first one should contain the table. The second page should contain a
  verbal description of the results of at most 150 words (one paragraph).
  You may use bullet points with one or two sentences each. The third page
  should contain the Stata code used for that question.
• The tables should look like those in the top finance journals (Journal of
  Finance, Journal of Financial Economics, Review of Financial Studies). In
  particular, the caption must define all variables and briefly explain what
  the table or figure shows. Tables and figures must be self-contained,
  requiring no further information from other sources, including the main
  text, to broadly understand them.
  Upload your final document using the appropriate link in Canvas. The document
  needs to be in pdf format.

Get Homework Help Now