Mixed Data Sampling Regression Research Articles

Forecasting China's GDP growth using dynamic factors and mixed-frequency data

Forecasting GDP growth is important and necessary for Chinese government to set GDP growth target. To fully and efficiently utilize macroeconomic and financial information, this paper attempts to forecast China's GDP growth using dynamic predictors and mixed-frequency data. The dynamic factor model is first applied to select dynamic predictors among large amount of monthly macroeconomic and daily financial data and then the mixed data sampling regression is applied to forecast quarterly GDP growth based on the selected monthly and daily predictors. Empirical results show that forecasts using dynamic predictors and mixed-frequency data have better accuracy comparing to traditional forecasting methods. Moreover, forecasts with leads and forecast combination can further improve forecast performance.

Density Forecasts with MIDAS Models

In this paper we propose a parametric block wild bootstrap approach to compute density forecasts for various types of mixed-data sampling (MIDAS) regressions. First, Monte Carlo simulations show that predictive densities for the various MIDAS models derived from the block wild bootstrap approach are more accurate, in terms of coverage rates, than predictive densities derived from either a residual-based bootstrap approach or by drawing errors from a normal distribution. This result holds irrespective of the data generating errors being normally independently distributed, serially correlated, heteroskedastic or a mixture of normal distributions. Second, in an empirical exercise we evaluate density forecasts for quarterly U.S. real output growth, exploiting information from typical monthly and weekly series. We show that the block wild bootstrapping approach, applied to the various MIDAS regressions, produces predictive densities for U.S. real output growth that are well-calibrated. Moreover, the relative accuracy, measured in terms of the logarithmic score or the continuous ranked probability score, improves for the various MIDAS specifications as more information becomes available.

Applications of MIDAS regression in analysing trends in water quality

Summary We discuss novel statistical methods in analysing trends in water quality. Such analysis uses complex data sets of different classes of variables, including water quality, hydrological and meteorological. We analyse the effect of rainfall and flow on trends in water quality utilising a flexible model called Mixed Data Sampling (MIDAS). This model arises because of the mixed frequency in the data collection. Typically, water quality variables are sampled fortnightly, whereas the rain data is sampled daily. The advantage of using MIDAS regression is in the flexible and parsimonious modelling of the influence of the rain and flow on trends in water quality variables. We discuss the model and its implementation on a data set from the Shoalhaven Supply System and Catchments in the state of New South Wales, Australia. Information criteria indicate that MIDAS modelling improves upon simplistic approaches that do not utilise the mixed data sampling nature of the data.

Unrestricted Mixed Data Sampling (MIDAS): MIDAS Regressions with Unrestricted Lag Polynomials

SummaryMixed data sampling (MIDAS) regressions allow us to estimate dynamic equations that explain a low frequency variable by high frequency variables and their lags. When the difference in sampling frequencies between the regressand and the regressors is large, distributed lag functions are typically employed to model dynamics avoiding parameter proliferation. In macroeconomic applications, however, differences in sampling frequencies are often small. In such a case, it might not be necessary to employ distributed lag functions. We discuss the pros and cons of unrestricted lag polynomials in MIDAS regressions. We derive unrestricted-MIDAS (U-MIDAS) regressions from linear high frequency models, discuss identification issues and show that their parameters can be estimated by ordinary least squares. In Monte Carlo experiments, we compare U-MIDAS with MIDAS with functional distributed lags estimated by non-linear least squares. We show that U-MIDAS performs better than MIDAS for small differences in sampling frequencies. However, with large differing sampling frequencies, distributed lag functions outperform unrestricted polynomials. The good performance of U-MIDAS for small differences in frequency is confirmed in empirical applications on nowcasting and short-term forecasting euro area and US gross domestic product growth by using monthly indicators.

A COMBINATION FORECAST FOR ENERGY-RELATEDCO2EMISSIONS IN THE UNITED STATES

This paper seeks to provide an alternative forecast to that provided by the Energy Information Administration (EIA) on energy-related monthly CO2 emissions in the United States. The data on CO2 emissions from petroleum, natural gas, coal and total fossil fuels obtained via the EIA covering the period January 2005 to November 2013 is analysed and then forecasted using ARIMA, Holt-Winters, and Exponential Smoothing prior to introducing the Singular Spectrum Analysis (SSA) technique for CO2 emissions forecasting. A new combination forecast (EIA-SSA) is also introduced by merging the SSA and EIA forecasts, and is seen outperforming all models including the EIA forecast. Finally, the EIA-SSA model is used to provide an alternative 12 month ahead outlook for US energy-related CO2 emissions from December 2013 to November 2014. This research is expected to influence the methodology adopted by the EIA for forecasting CO2 emissions in the future by improving the accuracy of the forecasts, and the impact of this study will be clearer upon comparing the actual CO2 emissions in US with the EIA, and EIA-SSA forecasts over the 12 month period which follows.

Mixed-frequency Cointegrating Regressions with Parsimonious Distributed Lag Structures

Parsimoniously specified distributed lag models have enjoyed a resurgence under the MIDAS moniker (Mixed Data Sampling) as a feasible way to model time series observed at very different sampling frequencies. I introduce cointegrating mixed data sampling regressions. I derive asymptotic limits under substantially more general conditions than the extant theoretical literature allows. In addition to the possibility of cointegrated series, I allow for regressors and an error term with general correlation patterns, both serially and mutually. The nonlinear least squares estimator still obtains consistency to the minimum mean-squared forecast error parameter vector, and the asymptotic distribution of the coefficient vector is Gaussian with a possibly singular variance. I propose a novel test of a MIDAS null against a more general and possibly infeasible alternative mixed-frequency specification. An empirical application to nowcasting global real economic activity using monthly financial covariates illustrates the utility of the approach.

Testing the functional constraints on parameters in regressions with variables of different frequency

We propose a test for the evaluation of statistical acceptability of a functional constraint which is imposed on parameters in the mixed data sampling regressions. The asymptotic behavior of the test statistic is characterized and a few other extensions are discussed.