Estimating Missing Observations in Economic Time Series

Harvey, Andrew; Pierse, Richard

doi:10.2307/2288346

Cited by 60 publications

(40 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In writing the program, considerable care was taken to devise a computationally efficient routine for evaluating the likelihood function. Particular attention was paid to the evaluation of PIIO, the matrix used to initialize the Kalman filter, and the algorithm adopted is described in some detail in our original research report (Harvey and Pierse 1982). Maximization of the likelihood function was camed out by one of the Gill-Murray-Pitfield numerical optimization routines in the UK NAG library, E04 JBF.…”

Section: Examplementioning

confidence: 99%

Estimating Missing Observations in Economic Time Series

Harvey

Pierse

1984

Journal of the American Statistical Association

Self Cite

240

148

View full text Add to dashboard Cite

show abstract

Section: Examplementioning

confidence: 99%

Estimating Missing Observations in Economic Time Series

Harvey

Pierse

1984

Journal of the American Statistical Association

Self Cite

240

148

View full text Add to dashboard Cite

show abstract

“…It can be adequately handled through a winsorized version of the observation y, to compute h,, so that the effect of extreme aberrant values is reduced. Alternatively, the masking effect can be investigated by treating an extreme outlier yT as a missing observation (for example Harvey and Pierse, 1984) and recomputing the hat matrix HIT]. Specifically, the masking effect of time T on time t -1 is assessed via hjT1lh,, i.e., the ratio of the statistics after and before "deletion" of the Tth observation.…”

Section: Discussionmentioning

confidence: 99%

Outliers detection in time series

Lee

Hui

1993

Journal of Statistical Computation and Simulation

View full text Add to dashboard Cite

This paper studies a diagnostic procedure for detecting outlying observations in univariate time series. The procedure takes the form of a graphical display of sample leverage with an envelope derived from simulation to assess the magnitude of outliers. The proposed method is more effective than techniques that based on sample residuals, especially when multiple/consecutive outliers are present. Practical illustrations using both simulated and business data are included.

show abstract

“…The methods available for generating forecasts for the missing observations include linear interpolation, OLS out-of-sample forecasts, forecasts based on the EÐ M algorithm and Chow and Lin (1976)'s BLUE estimations for the missing observations, among others. The approach adopted in this paper is to follow Harvey and Pierse (1984) who show that the recursive estimations of ARIMA models with the Kalman ® lter algorithm can produce consistent forecasts for the missing observations. First, each survey series with the missing observations omitted was subjected to the usual identi® cation process for the ARIMA models for ® nding an appropriate structure for each series.…”

Section: I D a T A D E Sc R I P T I O N Smentioning

confidence: 99%

Testing the rationality of exchange rate and interest rate expectations: an empirical study of Australian survey-based expectations

Kim¹

1997

Applied Economics

View full text Add to dashboard Cite

This paper examines the rationality of the Australian survey-based expectations, 1- and 4-week-ahead $US/$A exchange rate and 2 and 4-week-ahead Australian 90-day bank bill and 10-year bond rates. The actual and expected variables are found to be cointegrated, indicating that the expected future values and the future realizations of the exchange rate and interest rates have long-run equilibrium relationships. OLS estimations with the Newey - West corrections are employed for testing the unbiased expectations hypothesis (UEH) where the frequency of the expectations data is finer than the forecast horizons, and the exponential-GARCH models that take into account the time-varying nature of the forecasterror variance are employed for testing the weak rational expectations hypothesis (WREH). The evidence shows that the WREH could not be rejected in any case, except for the two-week-ahead forecast of the 90-day interest rate, which indicates that all available information is used at the time of forming relevant forecasts. The UEH, however, is decisively rejected in all cases. This indicates that strong rationality, which requires both UEH and WREH, is rejected in all cases. It is concluded that forecasters are weakly rational; however, their forecasts are not unbiased because the data available to them when forming expectations are inadequate.

show abstract

Estimating Missing Observations in Economic Time Series

Cited by 60 publications

References 0 publications

Estimating Missing Observations in Economic Time Series

Estimating Missing Observations in Economic Time Series

Outliers detection in time series

Testing the rationality of exchange rate and interest rate expectations: an empirical study of Australian survey-based expectations

Contact Info

Product

Resources

About