Feedback Pitching Documents 2022

Author

Stijn Masschelein

General Feedback

How to interpret the feedback

Most of my feedback in here is meant to help you write the proposal and eventually the thesis. So, if I highlight things that are not clear to me or if I have any questions that is something you can focus on in explaining better in the thesis.

General comments

  1. Don’t be hyperfocused on the data and the technical challenges. You want to make sure that you also explain the underlying theory or mechanism. This will help you to think about which control variables are necessary, which research design you can use, and what the take-away message is. This is the reason why I used explicit (small) theoretical models for the assignments. You need to have these theoretical1 models in mind when you interpret the data. This is why I want you to explain the mechanism or the economics better.

  2. Related to the point above. There are roughly two type of quantitative thesis topics. (1) You can investigate the effect of X on Y or (2) you can investigate how well you can predict Y. The first one is what I have focused on in my part of the course. Dirk focuses more on the second. Both have merit but I find that for an honours project, the second one is seductive in a bad way. It can lull you into believing that you can get away with just adding a lot of variables in a regression and hope for the best. The problem is that these regressions are hard to interpret 2, you have to care about out-of-sample prediction not just correct specification. Without a good theoretical understanding of what you are trying to do, it’s going to be really hard to interpret all you results.

  3. If you are relying on a script for webscraping, can you rely on an existing one? Do you have to write (some parts of) it? If it’s in R (or maybe Python) I can help. Especially, if the script will take a while to run, I can help you make it parallel. Your computer has multiple cores which all can do computations. Typically R runs on one core but sometimes it’s useful to run some code in parallel R sessions. Contact me in the R channel on Teams. Do not wait with starting on testing and writing the code.

  4. Don’t forget to define or describe your main variables. Especially if your main variables have a very specific meaning in the literature.

Chloe Truscott

  • Idea and Motivation: The underlying economics is not entirely clear. I think I prefer it better if you had less hypotheses and more about the mechanism. A dividend increase announcement seems to be bad for the peers but that is not really explained.
  • Data: I skimmed the JPE paper on peer measurement. Is the data freely available or will you have to construct it yourself? Why do you believe this is a better measure? Is it a better measure in general or will it be especially beneficial when you look at Dividend announcements.
  • So what: Do firms, still rely on dividends or are buybacks more relevant? I think the argument will not change if you replace dividends with buybacks.

Jackson Fletcher

  • Papers: All the key papers seem to be about “New Madness” not about the “Old Methods”.

  • Motivation: Don’t use technical terms (Total Value Locked) without introducing them3. It’s a bit the similar problem as with the papers. You are using the terminology of the “New Madness” not of the “Old Methods”.

  • Basic Hypothesis: Since TVL is a measure of interest in the protocol/token is the null hypothesis that there is no relation between interest and underlying value of the protocol/token?

  • All the independent variables seem to be indicators of prior interest in the token but not necessarily of the underlying value. To me this suggest more that you are trying to predict interest now from interest in the past. This is not a criticism. It just means that you probably are going to have to pay more attention to Dirk’s part of the course and that you need to be careful in how you interpret the results. You probably won’t be able to say that your independent variables cause the TVL/APY 4.

  • So What?:

    The market cap of the underlying token can differ from the market cap implied by the activity in the protocol, i.e the token is over/undervalued based on “fundamentals”.

    If this is true and you can explain it well, this would be really interesting. From your current writing, I am skeptical. It feels like you are making some implicit assumptions that I cannot tease out. Again, if you can explain this well in the thesis that would be really interesting. I think I would need to have more information on the economics: Why is the activity in the protocol not priced in the token? I thought that was the whole point of token based systems.

Qingyi Chen

  • Motivation, Idea: I want to get a better explanation of why a commitment to green projects would have an impact on the equity price of the firm and its peers.
  • Data: Which databases do you use for these variables?
  • Tools: I think I get from this description that you are planning an event study design around the issuance of peer green bonds.
  • Contribution: Is it really the bond market that is affecting the stock price of peers or is it the signal about the underlying business of the competitor that is driving the effect.
  • Random thought: Is a potential alternative explanation that a green bond issuance is access to cheap capital5 for the firm and that is what is reflected in peer reactions.

Chen Cui

  • Papers: Really good.
  • I like the motivation. One extra piece of motivation is that in a lot of studies that you cite, selection issues might be driving the result. Are female board members more risk averse or do risk averse firms attract more female board members? In your case, projects do not select the female entrepreneur.
  • Good simulation. Your simulated data does not have any selection effects for which you would need matching. Instead of round(runif(...)), you can use sample() with replacement. See my slides for lecture 2.
  • Data: Do you have the R script already for downloading the data or do you still have to write it?
  • You probably want to specificy country, category, and year as fixed effects. Now you are using for instance country as a numerical variable where there is the same effect if you go from country 1 to country 2 as if you go from country 5 to country 6.

Thomas Butler

  • Motivation: I would like to see a better description of what political uncertainty and the yield curve actually are. At least for the purpose of your research project. When you say “theory suggests”, it’s better to also give the mechanism. Which theory makes this prediction and why. You partly explain this in the Idea section where you refer to a move towards safe assets.

  • So What: How can you develop strategies for something that is unknown? (Not really important, just highlighting that you need to be careful in how you formulate the implications of your study).

  • Contribution: Since political uncertainty has already been studied in the equity market, do the effects run through the equity market or are they independent effects that would happen as well if we could keep the equity market consistent?

Philip Nguyen

  • Papers: I would like to see some Finance papers on the general effect of strict regulation on M&A behaviour.
  • Idea: I don’t think I get the argument. Stricter laws makes firms less profitable and therefor they are more likely to be taken over? What is the mechanism here? Why would a firm want to take over a firm that is not profitable. If there is such an effect would it not be more that some countries have laws that make it more difficult to operate in an environment with modern slavery 6. Enacting modern slavery legislation in an emerging market could then lift this burden.
  • Idea: The solution to the endogeneity problem that you are proposing is basically a difference-in-difference approach. Also, I am not 100% sure how your index could be affected by things that are not regulatory and legislative changes.
  • How do you measure deal attractiveness?
  • Good simulation. It does not incorporate any potential confounders. Just a warning.

Meizhu Chen

  • Motivation: Climate change is not the same as ESG. You can focus on a subset, that is completely fine. Just be aware that ESG is broader as a concern. I would personnaly expect that the effect of ESG incentives is hard to identify and I would not be surprised if your key papers do not have the best research design.
  • Idea: Why would mandatory reporting lead to increases in the use of incentives? What is the mechanism?

Joshua Laubsch

  • Key Papers/Motivation: What is geopolitical risk?
  • Motivation: This is more a random thought but to what extent is it surprising that risk is associated with volatility?
  • Idea: Why do you expect a difference in markets where producers are present compared to ones where they are absent?
  • Can you be more specific about how this would help with hedging? For instance in the run-up to the invasion of Ukraine what could investors have done if your hypotheses are supported. I am struggling a bit with what exactly we can learn from the study in addition to establishing an empirical fact. This would help me understand it better.

Henry Miao

  • Key papers: The papers seem to be focused more on cryptocurrencies than on valuation models.
  • Motivation: Is it the technical complexity that makes valuation hard or is it the immaturity of the markets?
  • Idea: What do you mean by “explain”? Do you mean predicting future price changes? I don’t think you can really identify the effect of these factors on prices. Are regression models the best way to determine these explanatory variables? Do you need to take into account possible non-linear effects? Do you take into account that this market is relatively new and that (the effect of) fundamental drivers can be quite different when cryptocurrencies emerge and where they are now.
  • What’s new: What do you mean by fundamental drivers? Variables like “active addresses” and “coin velocity” are probably partly driven by the price themselves. Which does not seem to be fundamental to me but that might be a language issue.

Peter Matthews

  • Cute title
  • Key papers:?
  • One issue that I see is that something else happened (Ukraine) to the energy markets in February 2022. Also, when was this decision made and or announced?
  • Motivation: Volatility implies increases and decreases, if you mean decreases I would just say decreases.
  • Data: Are there different price series available in India. It would be nice if you could find retail prices that are plausible less affected by the supply shock. This would help to adjust for other energy market effects.
  • The technical details for the ARIMA model are somewhat out of scope of the pitch. I did not put too much attention on it. I just want to acknowledge that you are using (a complex) event-study where you want to compare the current situation to the counterfactual where there was no supply shock. The question than is whether your timeseries model can credibly capture the counterfactual.

Footnotes

  1. but not necessarily formal and mathematical↩︎

  2. What does it mean to keep everything else constant↩︎

  3. I think I got it from the context↩︎

  4. Again, not a criticism↩︎

  5. interests are often higher than for regular bonds.↩︎

  6. see the Russia sanctions as an extreme example↩︎