Refine
Document Type
- Doctoral Thesis (1)
- Working Paper (1)
Language
- English (2) (remove)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- Zeitreihe (2) (remove)
Institute
We take a simple time-series approach to modeling and forecasting daily average temperature in U.S. cities, and we inquire systematically as to whether it may prove useful from the vantage point of participants in the weather derivatives market. The answer is, perhaps surprisingly, yes. Time-series modeling reveals conditional mean dynamics, and crucially, strong conditional variance dynamics, in daily average temperature, and it reveals sharp differences between the distribution of temperature and the distribution of temperature surprises. As we argue, it also holds promise for producing the long-horizon predictive densities crucial for pricing weather derivatives, so that additional inquiry into time-series weather forecasting methods will likely prove useful in weather derivatives contexts.
This thesis is concerned with the derivation of new methods for the analysis of nonstationary, cross correlated panels. The suggested procedures are carefully quantified by means of Monte Carlo experiments. Typical applications of the developed methods consist in multi-country studies, with several countries observed over a couple of decades. The empirical applications implemented here are the testing for trends in the investment share in European GDPs and the examination of OECD interest rates. In the first chapter, a panel test for the presence of a linear time trend is proposed. The test is applicable in cross-correlated, heterogeneous panels and it can also be used when the integration order of innovations is unknown, by means of subsampling. In the next chapter a cointegration test having asymptotic standard normal distributiun and not requiring exogeneity assumptions is derived. In panels exhibiting cross-correlation or cointegration, individual test statistics are asymptotically independent, which leads to a panel test statistic robust to dependence across units. The third chapter examines in an econometric context the simple idea of combining p-values from a series of statistical tests and improves its applicability in the presence of cross-correlation. The last chapter applies recent panel techniques to OECD long-term interest rates and differentials thereof, finding only rather week evidence in favor of stationarity when allowing for cross-correlation.