Research Project description Finance1 “Why have firms cash holdings increased lately?”
One of the big questions in corporate finance currently being discussed is the finding that corporate investment in tangible assets has substantially decreased in the past 30 years, while cash holdings have increased at the same time. The decline in investment is present even relative to investment opportunities (i.e. a decline in investment opportunities cannot explain it) and relative to cash flow or profits (i.e. changes over time of internal funds also cannot not explain it). In this exercise you are asked to replicate those findings on a dataset of the 100 largest US firms over the period 1980 to 2017.
More specifically, in the paper of Bates et al. (2009), that you are recommended to read (main understanding), the authors argue that an increase in volatility may be partly responsible for the increased cash holdings of firms. In Figure 2 of their paper, the authors show that firms operating in sectors with the highest growth in volatility (defined as the sector average standard deviation in cash holdings over a 10 year period) also show the highest increase in cash holdings. This suggests that increased volatility can at least partly explain the increase in cash holdings of firms. The aim of this project is to investigate to what extent the increase in cash holdings can be explained by an increase in volatility.
The two data files for this project contain (i) financial indicators of the 100 largest firms drawn from Compustat and (ii) a data file that has a firm-year volatility measure computed by taking the standard deviation in firms cash holdings over a 5 year period (minimum 2).
With these data files, students have to do the following:
- Merge the volatility data. Tomergethedata,checkwhichtwovariablethetwofileshaveincommon.Usethe merge1:1 command. For the syntax see clab 1.2 (Q 10).
- Generate two new variables inv (investment as percentage of book value) and cash (cash holdings as a percentage of book holdings). Before you do this ask yourself the following questions: a. Which variable represents investments?
b. Which variable represents book value
c. Which variable represents book value or book holdings?
d. Howdoyoucalculateinvestmentsasapercentageofbookvalue?Writedowntheformula. e. How do you calculate cash as a percentage of book value? Write down the formula
f. Look up the command to generate a new variable in clab 4.2 question 4 and plug in the formulas of d and e
- Showinagraphthatcashholdingsaspercentageofbookvalueofassetshasincreasedovertime, while investment as percentage of book value of assets has decreased at the same time; For this question, type:
collapse cash inv (semean) c_se = cash i_se = inv (count) n=cash, by(year) twoway (line cash year) (line inv year), name(gr1)
twoway (scatter cash year [aweight=1/c_se], ms(Oh) mc(blue)) /// (scatter inv year [aweight=1/i_se], ms(Oh) mc(black)) ///
(lfit cash year, lc(blue)) (lfit inv year, lc(black)), name(gr2)
4. In order to be able to interpret the constant, type: replace year = year-2000
5. Typesumcash_volifyear==0.Whatisthecashvolatilityintheyear2000?Astheyear2000isthe base year against which we compare all other years we also have to set the cash volatility in the year
2000 as the base. Type: replace cash_vol = cash_vol – r(mean) 1 Developed by Florian Peters and Tomislav Ladika.
- Show in two regressions what the average increase (decrease) in cash (investment) holdings is per year and test significance.
- Generate a variable which you call cash_vol_missing. This variable should indicate whether we have missing values for cash volatility. To generate a new variable look at clab 4.2 question 4 again. To indicate that something is missing you can write mi(variable) → the question you have to answer is: which variable do have use instead if the word ‘variable’? HINT: of which variable you would like to know whether a value is missing?
- Dotheregressionforcashwhileyouexcludethecompaniesforwhichyouhavenodataon volatility. This means that you have to write at the end of the regression: if cash_vol_missing ==x. Mind though that you still have to decide whether x should be replaced by zero or one. HINT: check your data file and look at the column cash_vol_missing if you don’t know the answer.
- Showthattheyearlyincreaseincashholdingsislowerwhencontrollingforvolatility.Exclude companies for which we don’t have any observations. This means that you have to use the same if-clause as in the previous question. What happened with the coefficients? What does it mean?
- Show that this result remains after removing the 5 largest values (outliers) of cash holdings and volatility. To find the outliers type: tab cash and tab cash_vol. Find the five largest values. Generate a new variable: gen insmpl = (cash< x) * (cash_vol<y). Fill in for x and y the cut-off points you found in the tab table (the lowest value of the five largest values). Show in a regression that cash holdings is still significant after removing these outliers while controlling for cash volatility while you exclude companies for which we don’t have any information on cash. At the end of the regression you write if ‘cash_vol_missing==x & insmpl==y’ . Find out what you should fill in for x and y. HINT: check your data file if you don’t know which number you should use.
- Interpret what percentage of the trend in cash holdings is due to volatility increase.
To interpret what percentage of the trend in cash holdings is due to volatility increase you can use your calculator.
- Finally, they have to describe these findings in a coherent paper.
The papers and data file are available for download on Blackboard.
Bates, Kahle and Stulz (2009). Why Do U.S. Firms Hold so Much More Cash than They Used to? Journal of Finance 64(5), 1985-2022
Gutierrez and Philippon (2017) Declining Competition and Investment in the U.S. NBER Working Paper 23583.
Jones and Philippon (2016) The Secular Stagnation of Investment? Working Paper.
Lee, Shin and Stulz (2015) Why Does Capital no Longer Flow More to the Industries with the Best Growth Opportunities? NBER Working Paper 22924.