Forecasting economic and financial time series with non-linear models

In this paper we discuss the current state-of-the-art in estimating, evaluating, and selecting among non-linear forecasting models for economic and financial time series. We review theoretical and empirical issues, including predictive density, interval and point evaluation and model selection, loss functions, data-mining, and aggregation. In addition, we argue that although the evidence in favor of constructing forecasts using non-linear models is rather sparse, there is reason to be optimistic. However, much remains to be done. Finally, we outline a variety of topics for future research, and discuss a number of areas which have received considerable attention in the recent literature, but where many questions remain.

Bootstrap prediction intervals for autoregressive time series

The calculation of interval forecasts for highly persistent autoregressive (AR) time series based on the bootstrap is considered. Three methods are considered for countering the small-sample bias of least-squares estimation for processes which have roots close to the unit circle: a bootstrap bias-corrected OLS estimator; the use of the Roy–Fuller estimator in place of OLS; and the use of the Andrews–Chen estimator in place of OLS. All three methods of bias correction yield superior results to the bootstrap in the absence of bias correction. Of the three correction methods, the bootstrap prediction intervals based on the Roy–Fuller estimator are generally superior to the other two. The small-sample performance of bootstrap prediction intervals based on the Roy–Fuller estimator are investigated when the order of the AR model is unknown, and has to be determined using an information criterion.

Medieval foreign exchange: a time series analysis

This chapter applies rigorous statistical analysis to existing datasets of medieval exchange rates quoted in merchants’ letters sent from Barcelona, Bruges and Venice between 1380 and 1310, which survive in the archive of Francesco di Marco Datini of Prato. First, it tests the exchange rates for stationarity. Second, it uses regression analysis to examine the seasonality of exchange rates at the three financial centres and compares them against contemporary descriptions by the merchant Giovanni di Antonio da Uzzano. Third, it tests for structural breaks in the exchange rate series.

The cyclical relations between traded property stock prices and aggregate time-series

This paper examines the cyclical regularities of macroeconomic, financial and property market aggregates in relation to the property stock price cycle in the UK. The Hodrick Prescott filter is employed to fit a long-term trend to the raw data, and to derive the short-term cycles of each series. It is found that the cycles of consumer expenditure, total consumption per capita, the dividend yield and the long-term bond yield are moderately correlated, and mainly coincident, with the property price cycle. There is also evidence that the nominal and real Treasury Bill rates and the interest rate spread lead this cycle by one or two quarters, and therefore that these series can be considered leading indicators of property stock prices. This study recommends that macroeconomic and financial variables can provide useful information to explain and potentially to forecast movements of property-backed stock returns in the UK.

Testing for unit roots in bounded time series

Many key economic and financial series are bounded either by construction or through policy controls. Conventional unit root tests are potentially unreliable in the presence of bounds, since they tend to over-reject the null hypothesis of a unit root, even asymptotically. So far, very little work has been undertaken to develop unit root tests which can be applied to bounded time series. In this paper we address this gap in the literature by proposing unit root tests which are valid in the presence of bounds. We present new augmented Dickey–Fuller type tests as well as new versions of the modified ‘M’ tests developed by Ng and Perron [Ng, S., Perron, P., 2001. LAG length selection and the construction of unit root tests with good size and power. Econometrica 69, 1519–1554] and demonstrate how these tests, combined with a simulation-based method to retrieve the relevant critical values, make it possible to control size asymptotically. A Monte Carlo study suggests that the proposed tests perform well in finite samples. Moreover, the tests outperform the Phillips–Perron type tests originally proposed in Cavaliere [Cavaliere, G., 2005. Limited time series with a unit root. Econometric Theory 21, 907–945]. An illustrative application to U.S. interest rate data is provided

Combined use of satellite estimates and rain gauge observations to generate high-quality historical rainfall time series over Ethiopia

Climate data are used in a number of applications including climate risk management and adaptation to climate change. However, the availability of climate data, particularly throughout rural Africa, is very limited. Available weather stations are unevenly distributed and mainly located along main roads in cities and towns. This imposes severe limitations to the availability of climate information and services for the rural community where, arguably, these services are needed most. Weather station data also suffer from gaps in the time series. Satellite proxies, particularly satellite rainfall estimate, have been used as alternatives because of their availability even over remote parts of the world. However, satellite rainfall estimates also suffer from a number of critical shortcomings that include heterogeneous time series, short time period of observation, and poor accuracy particularly at higher temporal and spatial resolutions. An attempt is made here to alleviate these problems by combining station measurements with the complete spatial coverage of satellite rainfall estimates. Rain gauge observations are merged with a locally calibrated version of the TAMSAT satellite rainfall estimates to produce over 30-years (1983-todate) of rainfall estimates over Ethiopia at a spatial resolution of 10 km and a ten-daily time scale. This involves quality control of rain gauge data, generating locally calibrated version of the TAMSAT rainfall estimates, and combining these with rain gauge observations from national station network. The infrared-only satellite rainfall estimates produced using a relatively simple TAMSAT algorithm performed as good as or even better than other satellite rainfall products that use passive microwave inputs and more sophisticated algorithms. There is no substantial difference between the gridded-gauge and combined gauge-satellite products over the test area in Ethiopia having a dense station network; however, the combined product exhibits better quality over parts of the country where stations are sparsely distributed.

The 30-year TAMSAT African rainfall climatology and time-series (TARCAT) dataset

African societies are dependent on rainfall for agricultural and other water-dependent activities, yet rainfall is extremely variable in both space and time and reoccurring water shocks, such as drought, can have considerable social and economic impacts. To help improve our knowledge of the rainfall climate, we have constructed a 30-year (1983–2012), temporally consistent rainfall dataset for Africa known as TARCAT (TAMSAT African Rainfall Climatology And Time-series) using archived Meteosat thermal infra-red (TIR) imagery, calibrated against rain gauge records collated from numerous African agencies. TARCAT has been produced at 10-day (dekad) scale at a spatial resolution of 0.0375°. An intercomparison of TARCAT from 1983 to 2010 with six long-term precipitation datasets indicates that TARCAT replicates the spatial and seasonal rainfall patterns and interannual variability well, with correlation coefficients of 0.85 and 0.70 with the Climate Research Unit (CRU) and Global Precipitation Climatology Centre (GPCC) gridded-gauge analyses respectively in the interannual variability of the Africa-wide mean monthly rainfall. The design of the algorithm for drought monitoring leads to TARCAT underestimating the Africa-wide mean annual rainfall on average by −0.37 mm day−1 (21%) compared to other datasets. As the TARCAT rainfall estimates are historically calibrated across large climatically homogeneous regions, the data can provide users with robust estimates of climate related risk, even in regions where gauge records are inconsistent in time.