Abstract
We propose new simultaneous and two-step procedures for reconciling systems of time series subject to temporal and contemporaneous constraints according to a growth rates preservation (GRP) principle. The techniques exploit the analytic gradient and Hessian of the GRP objective function, making full use of all the derivative information at disposal. We apply the new GRP procedures to two systems of economic series, and compare the results with those of reconciliation procedures based on the proportional first differences (PFD) principle, widely used by data-producing agencies. Our experiments show that (1) the nonlinear GRP problem can be efficiently solved through an interior-point optimization algorithm, and (2) GRP-based procedures preserve better the growth rates than PFD solutions, especially for series with high temporal discrepancy and high volatility.
Similar content being viewed by others
Notes
It should be noted that the state-space formulation of the PFD benchmarking technique provided by Quenneville et al. (2013) permits to coveniently deal with zero preliminary values. An extension to GRP benchmarking will be considered in the future research work.
In Di Fonzo and Marini (2012b) a Newton’s method with Hessian modification applied to a suitably reduced unconstrained transformation of the original constrained reconciliation problem is discussed. The interior-point method has proved to be much faster and reliable than the Newton’s method for reconciliation problems.
In contrast, feasible methods, like the Newton’s method discussed in Di Fonzo and Marini (2012b), have to start from a feasible point and then require a set of reconciled series as input. Given that deriving (preliminary) reconciled estimates to start the algorithm may be an additional complication for practitioners, we view this feature as another clear advantage of the IP method.
In general, for each low-frequency period the total number of constraints may differ: that’s why we use \(h_T\) to denote them. When this is not the case, it is \(h_T = \frac{r}{N},\, T=1, \ldots , N\).
Data and codes are available at request from the authors.
When the preliminary data presents a non-zero mean difference with the annual series, it is standardized to the overall level of the annual series according to the bias correction procedure described in Quenneville et al. (2009).
For each two-step procedure using ST at the second step, the obtained results are very close to those of the relevant simultaneous procedure.
References
Beaulieu JJ, Bartelsman EJ (2004) Integrating expenditure and income data: what to do with the statistical discrepancy? Tinbergen Institute discussion paper TI 2004–078/3. http://www.tinbergen.nl/discussionpapers/04078
Bloem A, Dippelsman R, Mæhle N (2001) Quarterly national accounts manual. Concepts, data sources, and compilation. International Monetary Fund, Washington DC
Bozik JE, Otto MC (1988) Benchmarking: evaluating methods that preserve month-to-month changes. Bureau of the Census-statistical Research Division, RR-88/07. http://www.census.gov/srd/papers/pdf/rr88-07
Brown I (2010) An empirical comparison of constrained optimization methods for benchmarking economic time series. In: JSM 2009 proceedings, business and economic statistics section, pp 2131–2143
Brown I (2012) Modernization of benchmarking economic time series at the U.S. Census Bureau. In: Federal Committee on Statistical Methodology 2012 research conference, Washington DC. http://www.fcsm.gov/12papers/Brown2012FCSMII-D
Causey B, Trager ML (1981) Derivation of solution to the benchmarking problem: trend revision. Unpublished research notes, US Census Bureau, Washington DC. Available as an appendix in Bozik and Otto (1988)
Cholette P (1984) Adjusting sub-annual series to yearly benchmarks. Surv Methodol 10:35–49
Cholette P, Dagum EB (1994) Benchmarking time series with autocorrelated survey errors. Int Stat Rev 62:365–377
Dagum EB, Cholette P (2006) Benchmarking, temporal distribution, and reconciliation methods for time series. Springer, New York
Denton F (1971) Adjustment of monthly or quarterly series to annual totals: an approach based on quadratic minimization. J Am Stat Assoc 66:99–102
Di Fonzo T, Marini M (2011) Simultaneous and two-step reconciliation of systems of time series: methodological and practical issues. J R Stat Soc C 60:143–164
Di Fonzo T, Marini M (2012a) Benchmarking time series according to a growth rates preservation principle. J Econ Soc Meas 37:225–252
Di Fonzo T, Marini M (2012b) New procedures for the reconciliation of time series. Department of statistical sciences, University of Padova, working paper series, n. 6. http://www.stat.unipd.it/ricerca/fulltext?wp=447
Di Fonzo T, Marini M (2013a) Benchmarking and movement preservation. Evidences from real-life and simulated series. In: Torelli N, Pesarin F, Bar-Hen A (eds) (2013) Advances in theoretical and applied statistics. Springer, Berlin, pp 449–459
Di Fonzo T, Marini M (2013b) A Newton’s method for benchmarking time series. In: Grigoletto M, Lisi F, Petrone S (eds) (2013) Complex models and computational methods in statistics. Springer, Berlin, pp 109–121
Helfand S, Monsour N, Trager M (1977) Historical revision of current business survey estimates. In: American Statistical Association, proceedings of the business and economic statistics section, pp 246–250
Nocedal J, Wright S (2006) Numerical optimization, 2nd edn. Springer, New York
Quenneville B, Fortier S, Gagnè C (2009) A non-parametric iterative smoothing method for benchmarking and temporal distribution. Comput Stat Data Anal 53:3386–3396
Quenneville B, Picard F, Fortier S (2013) Calendarization with interpolating splines and state space models. J R Stat Soc C 62:371–399
Quenneville B, Rancourt E (2005) Simple methods to restore the additivity of a system of time series. Paper presented at the workshop on ‘frontiers in benchmarking techniques and their application to official statistics’, Luxembourg, 7–8 Apr 2005
Stuckey A, Zhang X, McLaren CH (2004) Aggregation of seasonally adjusted estimates by a post-adjustment. www.uow.edu.au/craigmc/abs_agg_2004
Titova N, Findley D, Monsell BC (2010) Comparing the Causey–Trager method to the multiplicative Cholette–Dagum regression-based method of benchmarking sub-annual data to annual benchmarks. In: JSM 2010 proceedings, business and economic statistics section, pp 3007–3021
Trager ML (1982) Derivation of solution to the benchmarking problem: relative revision. Unpublished research notes, US Census Bureau, Washington DC. Available as an appendix in Bozik and Otto (1988)
Acknowledgments
A preliminary version of this paper was presented at the JSM 2012 of the American Statistical Association, San Diego, July 28th–August 2nd. We are grateful to Gary Brown, Irene Brown, Baoline Chen, Estela Bee Dagum, Vittorio Nicolardi, Neil Parkin and two anonymous referees for their helpful comments. We alone are responsible for the errors of the paper. The views expressed herein are those of the authors and should not be attribuited to the IMF, its Executive Board, or its management.
Author information
Authors and Affiliations
Corresponding author
Appendix: Gradient and Hessian of the global GRP criterion
Appendix: Gradient and Hessian of the global GRP criterion
In this appendix we present the analytical expressions of the gradient vector and of the Hessian matrix for the function \(F_{\textit{GRP}}\) as defined in (5), which can be exploited by Newton-type nonlinear programming optimization procedures. The derivation is a straightforward extension of the expressions shown by Di Fonzo and Marini (2013b) for the univariate GRP criterion.
Let us denote with \(y_{j,t}\) and \(p_{j,t}\), respectively, the \(j\)th target and preliminary series of the system observed in the period \(t\), with \(j=1,\ldots ,m,\, t=1,\ldots ,n\), where \(m\) is the number of variables and \(n\) the number of the high-frequency periods. The value \(\{y_{j,t}\}\) is the \([i+(j-1)n]\)th element in the stacked \(nm\)-dimensional vector \(\mathbf{y}\). The gradient vector of \(F_\textit{GRP}\) is the \((nm \times 1)\) vector
where
for \(j=1,\ldots ,m\) and \(t = 2, \ldots ,n - 1\).
Let us denote the elements of the Hessian matrix, \({\nabla }^2 F_{\textit{GRP}}( \mathbf{{y}} ) = \mathbf{H}(\mathbf{y})\), as
Notice that the Hessian matrix is both symmetric and tri-diagonal, that is its non-zero items are \(h_{s,s},\, s=1,\ldots ,nm,\, h_{s-1,s},\, s=2,\ldots ,nm\), and \(h_{s+1,s},\, s=1,\ldots ,nm-1\). After some calculations, for \(j=1,\ldots ,m\) we find:
Rights and permissions
About this article
Cite this article
Di Fonzo, T., Marini, M. Reconciliation of systems of time series according to a growth rates preservation principle. Stat Methods Appl 24, 651–669 (2015). https://doi.org/10.1007/s10260-015-0322-y
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10260-015-0322-y