(4). ECONOMETRICS
Questions about the topics within the subject area of Econometrics should be directed to: Christian Møller Dahl, Department of Business and Economics, e-mail: cmd@sam.sdu.dk
All topics in this subject area can be written in English or Danish.
4.1 Inequality in life quality, health and health behavior – a geographical perspective (Jørgen T Lauridsen)
Denmark is known to be one of the most equal countries in the world when it comes to income. Therefore, it is quite intriguing to realize that the social inequality in health is very high in Denmark. While this topic has been investigated extensively in literature, geographical aspects alike potential geographical inequality has been much less investigated. To our knowledge, no such studies considered the Danish case.
The purpose of the suggestion could be to investigate inter alia:
- To which extent is there geographic inequality in life quality, health or health behavior
- To which extent may such inequality be ascribed to geographic variation in economy, demography, policy etc?
- Are the present policy initiatives appropriate?
The empirical investigation can be based on a rich source of data, aggregated to the municipal level.
The project can be written in Danish or English. However, a basic ability in reading Danish is highly recommendable, as some of the data sources and references are in Danish.
4.2 Social inequality in health – can it be measured, and how? (Jørgen T Lauridsen)
Denmark is known to be one of the most equal countries in the world when it comes to income. Therefore, it is quite intriguing to realize that the social inequality in health is very high in Denmark. While this topic has been investigated extensively in literature, the empirical approaches have been quite rudimentary in certain dimensions: Income has generally been used as a measure for socioeconomic status, even though severe endogeneity may exist between income and health; health has been measured using simple and subjective measures like self-assessed health, the development over time is less investigated etc.
The purpose of the suggestion could be to investigate inter alia:
- Can we suggest and apply improved measures of social status?
- Can improved results be obtained by using objective health measures?
- What happens over time with health inequality in DK?
- Are the present policy initiatives appropriate?
The project may conveniently be carried out as an empirical investigation, based on the Survey of Health, Ageing and Retirement in Europe (SHARE), which enables comparison across countries, but different Danish sources are also available.
The project can be written in Danish or English. However, a basic ability in reading Danish is helpful, as selected relevant references are in Danish.
4.3 Big data visualization and predictive analytics (Christian M. Dahl)
This project is about big data visualization and predictive analytics applied to one of the ongoing competitions hosted by www.kaggle.com.
The amount of available data in organizations (private and public) is growing exponentially. But more data doesn’t automatically translate into information that is useful and facilitate better decisions. Unfortunately, the capacity of most organizations to analyze data has not increased at the same pace as the available data. To replace gut feeling based on experience with a data-driven approach we need to enhance this capacity by introducing visualization and predictive analytics.
Big data visualization and predictive analytics, that is based on the fundamental principles of statistics (econometrics), trains a computer model to automatically learn from large amounts of data to find the complex, hidden patterns that can optimize your investment in financial assets; your inventory; predict fraud, maintenance, or customer retention; recommend the products that customers actually need; or even diagnose Alzheimer’s disease.
Big data visualization and predictive analytics has gotten a lot of attention recently through the success of Kaggle. Kaggle is a web platform where organizations like General Electric, Pfizer and Facebook host predictive modeling competitions (based on a wide range of different data sources) with prices up to $3M.
4.4 Realized Variance for Cryptocurrency Predictability (Christian M. Dahl)
Exploring the predictive power of estimators of the equity variance risk premium and the conditional variance for cryptocurrency returns. The Realized Variance (RV) has been a major focus of research into accounting for uncertainty in financial investments. The objective of this project is to compare different RV forecasting models and then applied the best forecast to predict one week-ahead Bitcoin returns. In the absence of intra-day data from cryptocurrencies, the student is welcome to use daily prices from https://coinmarketcap.com/coins/ . The RV will be calculated with a weekly frequency.
4.5 Estimating the effect of policy interventions using the Synthetic Control Method (Giovanni Mellace)
The Synthetic Control Method (SCM) allows estimating the effect of a policy intervention in setting where a panel of aggregated data on few units is available.
Only one (ore few) “treated unit” is affected by the intervention while the other units are used to mimic what would have happened to the “treated unit” in the absence of the intervention.
This is done by finding a linear combination of the pre-intervention outcomes of the non-treated units which is as close as to the pre-intervention outcome of the treated unit in each period.
SCM is very easy to implement and a very well documented package is available in R.
A suggested project could be using SCM to re-evaluate the effect of Democracy on growth.
4.6 Causal Machine Learning (Giovanni Mellace)
Machine learning technique like random forests and neural networks are becoming popular in all field of economics.
Although this methods are great tools for prediction can only help to a limited extend to uncover causal effects.
The literature on Causal Machine Learning has been growing exponential in the past years and is becoming prevalent in applied work in economics.
Both theoretical and more empirical projects are possible in this field.
4.7 Applying the Difference-in-Differences method to study the impact of antibiotics on society (Volha Lazuka)
Some interventions, such as antibiotics, usually arrive sharply and spread quickly across the country.
Yet, you can estimate the causal effect of these interventions by using both a time dimension and a region dimension such as pre-intervention level of disease.
This implies the use of Difference-in-Differences (DD), the most popular quasi-experimental method, with either binary or continuous treatment.
While a canonical DD does not require inclusion of covariates, one may be willing to use them to improve the identification assumptions.
There is a special estimation procedure how you should implement DD with covariates.
R package for DD with and without covariates is available and straightforward to use.
4.8 Combining time series and panel data to study the impact of disease or income shocks on society (Volha Lazuka)
Detrended time series of mortality and price variations can be used to identify time-varying negative shocks that in certain contexts can be seen as causal.
These contexts include developed countries in the past and developing countries nowadays.
Such negative shocks have been found to have a profound effect on various demographic outcomes of the society, such as mortality, fertility and migration.
You can study the impact of these shocks in different models, including linear and non-linear, such as duration models.
A suggested project could be using detrended time-series and duration models in R in the historical context. The use of developing context is also possible.
4.9 Modelling Seasonality in Quarterly and Monthly Time Series (Nils Karl Sørensen)
Time series at the quarterly or monthly frequency is likely to exhibit seasonal behavior. Such series may be med modelled with a dummy approach, but in case of non-stationarity a model with unit roots are preferred.
A BA-project could take point of departure in the literature by Hylleberg, Engel, Granger and Yoo (quarterly case) or Phillip Hans Fransens (monthly case), and then focus on testing data for seasonal unit roots at the seasonal frequencies.
Applications could be on OECD national accounts data, and then look for example at common seasonal features or on data related to tourism nights.
4.10 Estimating Monetary Policy in an International Perspective (Nils Karl Sørensen)
Monetary policy is global and especially the USD, EURO and Yen are steering the marked relative to the interest rate and also to the money demand.
A BA-project could take point of departure for example in the quantitative easing monetary approach and use quarterly data from the Millennium to present to estimate model of the monetary efficiency in a purchasing power parity set up.
Data could be collected at the home pages of the centrals banks and model set up could focus of short and well as long run implications of the policy. Models could be estimated as single models as well as in a jointly set up.