First, you need to install matlab before taking my lectures because I explain everything using matlab codes. BTW, you can use R in case you can use it (Rapach provides R codes in his website. Most of stuffs I am going to teach can be done with these codes) but I will provide only limited support on R.
Second, if you donâ€™t have any access to Database, then that will be really issue. It turns out you may not be able to get any data from Bloomberg. Anyway, in that case, could you search for Datastream tickers of your predictors? For example, you are going to say you need short term interest rate for your chosen country and need to tell me the ticker for that is xxxx. Then I will be able to collect data xxxx for you.
I would like to provide some info. on datastream ticker (called Mnemonic). You need first search for Mnemonics of aggregate market return/risk free rate for your chosen country. I am attaching some references you can use when you try to locate these.
At least you have an access to WRDS. For example, you can get an excess return to the market index from MSCI available in WRDS. I think you could find some useful data in WRDS while you still need to ask me to get most of data from datastream. BTW, if you want to get macroeconomic variables (e.g., interest rate and exchange rate, etc), you could use data service from IMF/OECD/US FED (Saint louis and FED BOARD). These are free.
I have posted the first video clip on lecture slide 1 and one matlab code. This is available under the folder â€˜Return Predictabilityâ€™ inside â€˜Week 9 and 10: Regime Switching Modelsâ€™. BTW, please read all of my emails from June 2. Some of you havenâ€™t responded to my email at all. I am a bit worried. Please let me know if you have any problem.
I would like to confirm several things (For more info. on the following things, you need to read my previous emails).
1) Program: you should have matlab in your pc.
2) Data: you should have an access to WRDS (please check my previous email on how to get it). And if you need certain data series for your dissertation, you could find the ticker (mnemonic) of that for Datastream (please see datastream.zip I sent for some info.). If you provide me that information, then I could get the data for you. For macroeconomic variables, you could use IMF/OECD/US Fed (saint louis)
3) please read my previous emails on how to search for predictors in your chosen country.
4) You also need to provide me a list of recent papers on return predictability on your chosen asset by the end of this week.
I have uploaded all lecture slides/matlab codes you need. Please notice that I have changed the name of matlab codes in lecture 2 folder (perhaps it would be easier if you download all files again).
I will post my next video clips within a few days. BTW, you need to collect the returns to your chosen asset and risk-free rates for your country. I have one announcement on this. You will find both $ returns and local currency returns for your chosen asset. Normally you will use local currency market return and risk free rate. Here the assumption is that you are an investor in that country, and want to invest there. On the other hand, if you use $ market return and $ risk-free rate, that means you are a us investor and want to invest into that market. Now, in this case, you have to deal with what would impact the exchange rate of US $ against the local currency (since $ return includes the changes in exchange rate), and it involves with bunch of exchange rate related theory/story/variables. So, I think it would be more appropriate for you to use local currency asset return and local currency risk free rate.
BTW, if you want to know 1) how to generate technical indicators, 2) how to generate dividend price ration, and 3) how to get the US variables used in Welch and Goyal by using the raw data in Goyalâ€™s website, you can look at the matlab codes I have uploaded in â€˜searching for predictorsâ€™ holder. I will explain this later but you are ok to try running this if you want.
I have posted two video clips on BB. The first is about lecture slides 2,3,4 and class02program_2019.m, and the second is about matlab codes generating US predictors (using Goyalâ€™s raw data), technical indicators, and dividend price ratio. If you have any questions, ask them on the discussion forum as I explained in my previous email. I will post more video clips next week.
I have two announcements. First, clearly I have no intention to cover everything in Rapachâ€™s lecture slides. His slides are for 1 semester teaching (3 to 4 months). So, I select/teach only relevant (arguably most important) parts of his slides for your thesis, and accordingly my lecture video clips only cover them. For your thesis, you need to study only the material I cover in my video clips.
Second, I would like to provide some info. on how to structure your MSc dissertation. Typically there are 6 or 7 sections in your thesis in the following order. I am attaching an exemplar MSc thesis as a reference.
1. Introduction: you need to provide some â€˜motivationsâ€™ at least briefly such as why return forecasting is important, why you chose your country, etc. You can also provide a brief overview of your thesis.
2. Literature review: first you need to summarize at least key papers on stock return forecasting (papers on US and international predictability). And you should have a separate section summarizing
a list of predictability papers using your countryâ€™s data.
3. Methodologies: you need to summarize all the techniques used in the thesis clearly as much as you can. For example, any methodology you learn from me should be explained in details as much as you can (for this,
you may need to provide more details than the materials in the slides by reading some textbooks on them).
4. Empirical results: I expect you to write the following.
â€“ Data section: provide definition/source of data you use. You need to provide descriptive statistics for all variables (mean, variance, skewness, kurtosis, autocorrelation, etc)
â€“ In-sample regression results: regression results on the whole sample. We will talk about this in the last lecture.
â€“ Out-of-sample forecasting results: everything you have learned and decided to use based on the lecture materials.
You need to study all of my lecture slides to know what you are going to do in â€˜in-sampleâ€™ and â€˜out-of-sampleâ€™ sections. You need to report/summarize every empirical results.
5. Conclusion: your verdict on return predictability on your chosen asset, and discuss implications and limit on your research (along with some discussions on possible future research).
7. Appendix (if necessary)
Note 1: you can put all the figures/tables at the end (before Appendix) if you want.
Note 2: when you write equations and provide tables/figures/references, you should do it consistently and clearly. Please read the exemplar thesis for more information.
I have posted two video clips in â€˜lecture 2â€™ folder. I will post two more video clips (lecture 8 and 9) by the end of this week.
You should have now attempted to collect market index return, risk-free rate, and some predictors for your chosen country. If you havenâ€™t, please try doing it this week.
Once you compute excess return and collect some predictors, you should try modifying the matlab codes with your own data. You can ask questions on discussion forum or by email in case you have difficulty in understanding my lectures/matlab codes and data collection/code modification.
Most of you have sent me the list of predictors for your chosen asset. Thanks for that. I think some of you havenâ€™t read one of my emails on this carefully. I will provide that again with some more comments. Here is what I wrote before + new comments (in red)
I have recommended you to search for possible return predictors for your countryâ€™s market excess return. For that purpose,
I would like to provide more information on how to do that. Inside the folder â€˜searching for predictorsâ€™, you can find the following information.
Please read the papers mentioned below, which I have posted in this directory, and study the video clip on â€˜searching for predictorsâ€™ I posted.
1) Typical Predictors: this is part of Goyal and Welchâ€™s paper on typical return predictors for US market excess returns.
You donâ€™t need to search for the updated data. You can find this in Goyalâ€™s website.
If your country is not U:S, then you need to search for similar variables in your chosen country (via Bloomberg or others). Be aware of this. Some variables are only
available on US, but not in your country. BTW, US predictors can be still useful if your chosen country is not US. See 3) below.
You should aim to find the corresponding variables used in Goyal and Welch for your country. But only some of those will be available. I commented a bit in my video clip in â€˜searching for Predictorâ€™ folder as follows. Please listen to it. I think you should be able to find at least dividend price ratio, short-term interest rate, default spread, term spread, inflation, stock market volatility.
2) Technical indicators: One of Rapachâ€™s papers demonstrates that technical indicators (e.g., simple trading rule based on return/volume signal) can forecast US market excess return.
Perhaps you can say the same thing to your country. Please look at what are the necessary data to generate this.
You can use a group of equivalent technical indicators used in Rapachâ€™s paper for your country.
3) Role of US variables: One of Rapachâ€™s papers demonstrates that US variables could be helpful to forecast other countriesâ€™ market excess returns.
Some of the US return predictors or market excess returns in 1) (available at Goyalâ€™s website) can be useful.
So, it is possible to include US variables as one of potential predictors for your non-US asset.
4) Investor sentiment: this is one of newly discovered predictors for US. If you can find similar one in your chosen country, that will be great.
But this might not be possible. Because of 3), you could still use US investor sentiment for your chosen country.
Search for investment sentiment data for your country. Investment sentiment data for US is available in Zhou (one of the authors of the investment sentiment paper)â€™s website.
5) Various factor premiums: One of Rapachâ€™s new papers claim that factor premiums (e.g., SMB, HML, etc) can forecast US market excess returns.
Perhaps this would be true to other countries. Or because of 3), US factor premiums can be useful for other countries.
And perhaps you can find some relevant factor premiums for your chosen country in Frenchâ€™s data library (just google it).
You can find the data in Frenchâ€™s data library (this is French in Fama and French model)
6) I have added a recent survey on international return predictability there. Perhaps you can find some info. from reading this paper.
And please also search for some unique predictors used for your chosen country from the literature. Please read this.
I think you need to decide which variables should be used in your dissertation. Often data availability will determine that. Please make a list of
predictors you wish to use in your dissertation based on data availability and your preference. And please try to search for these variables, and
once you determine which variables to include, please let me know about that.
You need to add unique predictors you have located from the literature to the above lists. And it is your decision to put which variable is used in your dissertation. Any variable, which you can justify why that is a useful predictor for your chosen asset, and you can find, can be included. I guess at least more than 10 predictors should be chosen for your thesis.
I have a couple of announcements on data collection.
1) At least you need to find datastream tickers (mnemonics) for market index return and short-term interest rate (see â€˜Datstream.zipâ€™ I sent for some info. on mnemonics) very quickly.
2) It is completely up to you which predictors are used in your dissertation. However, at least you need to use dividend price ratio and short-term interest rate in your country (see Rapachâ€™s papers on the role of US in international return predictability I posted on BB). For non-US countries, dividend price ratios might not be available. Then you need to use total return index/price index for your market index to construct it. This is explained in â€˜searching for predictorsâ€™. And even finding short-term interest rate might not be easy for your country. You need to read the data section of the return predictability paper you found to see how others cope with this issue. I have also provided some information inside â€˜Datastream.zipâ€™.
3) If you use a group of US predictors for your country, you can find most (or all?) of these without my help. See the instructions I wrote in â€˜searching for predictorsâ€™ on BB for more information.
4) Once you make a list of potential predictors for your chosen asset, you need to search for mnemonics for them. You also need to check if any information is available from the return predictability papers you have found on your asset. In many cases, perhaps equivalent data series (used in US return predictability) might not be available for your country. Data availability is one of big issues in international return predictability. Without mnemonics, I canâ€™t find the data from datastream. I have only limited access to datastream because of weird errors in datastream with excel in windows 10 system.
5) Try WRDS to see if some variables available there could be used in your research. For example, some macro/financial uncertainty data are available there.
6) You can find some macroeconomic variables easily in Fed (Saint louis), IMF, OECD such as consumer price index (for computing inflation) and GDP (for computing gdp growth). You donâ€™t need an access to Datastream for this.
My expectation is that you need to start estimating your models (by modifying the matlab codes you have) at least from July 1st. So, at least you should be able to compute excess returns for your asset and collect a group of potential predictors (at least US variables even in worst cases) before that. Once you have those, at least you can try modifying/estimating all the codes I have provided with these. Once you have good codes, then you will be able to quickly run the same codes with more variables later if you want.
I will help you a bit more on finding market return and related variables for your chosen index. I have done this for Antonios for his request and have decided to do the same for others since some of you seem to have a great difficulty.
Can you send me the name and tickers (if possible) of the market index you use (in Datastream)? Then I will provide price index/total return index/dividend yield/price earning/price-book/market value related to that market index.
BTW, please read also the data section of past return predictability papers on your chosen asset to find out the data source/list of variables used in the literature. That could provide some information on how to get the data for you.
Finally, I would like to share Antoniosâ€™s list of predictors for his chosen asset and his summary of the variables/sources and his effort to find them on the web. I think what he did is excellent (very close to what I asked you to do), and all of you should try doing the similar things unless you have your own plan.
Once you make the list in this way, the next step is to try your best to find the data on the web and identify the missing variables, and finally ask me if I can locate them from Datastream. The problem is that sometimes data might not be available for your chosen country. That is just a reality of the empirical investigation when we use non-US data.
Dependent variable: STOXX Europe 600
Â· Potential problem: Small number of observations available in the data sets found; especially in monthly framework. For EUROSTOXX 50 I found relatively larger time series of data starting from 1/1987 provided by ECB but without volume observations.
Â· For STOXX â€“ datastream ticker STXPE
Â· ECB short term interest rates (EIBOR3M- 3 month Interbank rate) and (EUROMLR-Marginal Lending rate), also provided by ECB
Â· GDP growth Europe-OECD
Â· Long term yield of German Bond-FRED Economic data
Â· Europe Default rate (Spread between corporate bonds and German government bond) â€“ Bundesbank Website
Â· Consumer Price Index-European central bank
Â· Economic sentiment indicator-European central bank
Â· European Sentix investor confidence-European Central Bank
Â· Spread between 10y German bond and peripheral, high debt countries bond FRED.Economic data
Â· US interest rates capture by (FRTBS3M-3)-Also provided by Fed
Market Related Variables
Â· SMB-European Markets- (From French data library)
Â· HML-European Markets
Â· RMW-European Markets
Â· CMW-European Markets
Â· Log Equity risk premium volatility-European Markets
Â· Equity risk premium volatility-European Markets
Â· Dividend Yield ER(DY) Aggregate ER index from Datastream if there is not STOXX Eur. 600 dividend data
Â· Market Value( ER(MV) to calculate the lagged price dividend yield)
Â· Earnings price ratio (ER(EP))
Â· Default yield spread (STOXX Europe 600-Spread between junk bonds and high rated bonds)
Â· Moving average (1,6) Signals
Â· RSI (6)
Â· On Balance Volume (VO if volume data available for STOXX Europe 600)
As promised, I have posted two video clips (lecture 8 and 9) in â€˜lecture 2â€™ folder. I will post two more video clips (â€˜lecture 11â€™ in lecture 2 folder and â€˜in-sampleâ€™ in lecture 3 folder) by the end of next week. No more clips after this!!
I am really worried. I have just googled and find two references on return predictability in China stock market. Could you read the following papers as a starter for your thesis?
Can US Economic Variables Predict the Chinese Stock Market? https://www.sciencedirect.com/science/article/pii/S0927538X12000777
How Predictable Is the Chinese Stock Market?
Perhaps you could just try updating the above two (e.g., by using same variables but with recent data) in your thesis.
I am sending price index/volume of three indices (Shenzhen composite, Shenzhen A share, Shenzhen B share). Only these two are available for these indices (not price dividend ratio, etc) While it is unfortunate not to have total return index, you can at least use these price indices to compute a proxy of market returns (I donâ€™t know which index is preferable, so I send these all). You need to check which one is the best proxy from past papers. I think you can also compute technical indicators with this. The data series are from 1992 or 1993.
And I also send short-term interest rate with longest data span (from 1993). I think you can compute the market index return based on these. And you can also use this as one of predictors.
Then you could try finding some macroeconomic data (e.g., inflation and gdp growth) from IMF and add those to the list of predictors and perhaps a group of US predictors could be also used. I would say this is much less riskier than choosing to predict US market excess returns.
The post install matlab before taking my lectures because I explain everything using matlab codes. appeared first on Best Custom Essay Writing Services |Essaytreatise.com.