Econometrics Assignment
1. Clinical Cancer Trials, Part I
In their paper "Do Firms underinvest in Long-Term Research? Evidence from Cancer Clinical Trials", Budish et al. (2015) analyze whether private research investments are distorted away from long-term projects in the context of cancer studies.
The authors use data on 80 different types of cancer and three different stages (localized, regional, metastatic) for patient cohorts from 1973 to 2004. The observations are on the stage-level. The subscript cs indicates cancer stage. In other words, Mcs measures variable M for cancer type c and stage s.
The paper is particularly interest in how the structure of the patent system affects research investments:
Patents award innovators a fixed period of market exclusivity (e.g., 20 years in the United States). Yet, since in many industries firms file patents at the time of discovery (\invention") rather than first sale (\commercialization"), effective patent terms vary: inventions that commercialize at the time of invention receive a full patent term, whereas inventions that have a long time lag between invention and commercialization receive substantially reduced or in extreme cases, zero effective patent terms. [...] The fixed patent term reduces the number of calendar years for which private firms enjoy monopoly protection on investments, and excess discounting reduces the weight the private firm places on each of those years relative to the societal weight.
Budish et al. investigate this issue in the context of cancer research. The key reason for focusing on the cancer market is that there is a good predictor of how long it will take to commercialize a drug: survival time. To get FDA approval, firms have to demonstrate that the drug is safe and effective, where \effective" is usually interpreted as improving survival.
The authors thus conjecture a negative relationship between clinical trial activity and survival rate.
1. The dataset cancer stage.dta contains the data used in the paper. Load the dataset into STATA and get an overview over the dataset (STATA: describe and sum)
2. Consider the model in equation (10) on p.2067. Scs is the main independent variable of interest. Xcs is a vector of covariates, the corresponding coefficients are contained in the vector λ.
(a) In your dataset, which variables do Ycs and Scs correspond to?
(b) Estimate the following model by OLS:
log(Ycs) = α + βScs + ∈cs
(c) Interpret β^.
3. Market size is widely considered to be an important determinant of R&D expenditures. Budish et al. therefore control for market size in some of their regressions.
(a) How is market size measured in the paper? (1 sentence)
(b) Generate the variable market size as described in the paper.
(c) Estimate the following model by OLS:
log(trials) = α + βsurv5yr + λ1 log(market size) + ∈
(d) Is the coefficient estimate λ^1 significantly different from 0? Formally state the null and the alternative hypothesis and conduct the test on the 1% level. Conclude (1 sentence).
(e) Omitted variable bias:
i. Compute the OVB of the OLS estimator of β when market size is not included in the model.
ii. Using an appropriate auxiliary regression, name and compute the two components of the OVB.
2. Clinical Cancer Trials, Part II
Consider the following statement:
If heterogeneity in demand for treatments or a paucity of scientific opportunities were driving the survival time-R&D correlation, the observed correlation should be independent of whether surrogate endpoints are used.
1. Describe what surrogate endpoints are (max. 3 sentences)
2. Surrogate endpoints are more often used for hematological malignancies.
(a) Generate the interaction term and estimate the following model by OLS:
ln(Ycs) = α + βScs · Hc + γHc + δScs + ∈cs
where Hc is a dummy variable for hematological malignancies. In our dataset this variable is called blood.
(b) Do your estimates confirm the above statement? Your answer must be based on a formal statistical test at the 5%-level.
Attachment:- Assignment File.rar