Review of Financial Studies, 17 (2004), 1-35.
We develop and analyze a
model of a multi-stage investment project
that captures many features of R&D ventures and
start-up
companies. An important feature these problems share is that
the
firm learns about the potential profitability of the
project
throughout its life, but that technical uncertainty about the
research and development effort is only resolved through
additional investment by the firm. Consequently, the risks
associated with the ultimate cash flows the firm realizes on
completion of the project have a systematic component even while
the
purely technical risks are idiosyncratic. Our model captures
these different sources of risk, and allows us to study
their
interaction in determining the risk premia
earned by the venture
during development.
We show that the systematic risk,
and the required risk premium, of the venture are highest
early in
its life, and decrease as it approaches completion.
To download a PDF file with the closed form solution in this case, click here
Caveat Emptor --- The code used in the paper is available below although we make no promises about readability.
The tables and figures in the case without learning were generated in a two step process. First a c program with source file genconst.c generates the constants for each stage in the closed form solution. You will need a number of Numerical Recipes in C routines. For obvious reasons I have not included them. If you do not already have them, you will need to purchase them. This program takes input.tex as an input file (yes, this file is TeXable). The constants are then read out into files that the mathematic spreadsheet constants.nb reads in and generates the figures and tables. Click here to download a self extracting archive, or here for a zip file that contains all these files.
Things are much harder for the case with learning because we do not have a closed form solution. First a c program with source files nfppde.c and win.c (this second file may or may not be required depending on your compiler) generates an output file that contains the value and risk premium over a range of values specified interactively at the outset. The input file that specifies the parameters and accuracy is called input.tex (and is TeXable). Once this program has run (and it can take a while), the output files are read by the mathematica spreadsheet out.nb. All graphs and tables in the learning section of the paper are generated by this spreadsheet. Click here to download a self extracting archive, or here for a zip file that contains all these files.
IMPORTANT NOTE: The program is written so that alpha and beta in the program relate in the following way to a1 and a2 in the paper:a=a1 and b= M a2- a,where M is defined in Appendix C. So, for example, if M=4, a1=1 and a2=0.5 (as it is in the paper) then a=1 and b = 4 0.5 -1 =1.