Summary of Prof. Espasa’s contributions to the field of economic forecasting
Espasa and Pérez (1979), which was not published because it went against Bank of Spain policy at the time, developed a series of quantitative models for the instrumentation of monetary policy, for the first time in Spain. These models were systematically used by the Bank of Spain until the institution changed its control strategy. Espasa and Salaverría (1988) performed an in-depth review of the initial econometric models.
Espasa (1980) represented the first study of the demand for cash in the Spanish economy and Escrivá and Espasa (1988) developed and estimated a model concerning the determination of reserves in the banking system.
Espasa and Cancelo (1987) estimated a model to predict daily fiduciary currency circulation demand.
Publications in international academic journals and books
- Escrivá, J.L. and A. Espasa, (1988), “An econometric model for determination of banking system excess reserves”, chapter 22 in Economic Modelling in the OECD countries, edited by H. Motamen, Chapman and Hall.
- Espasa, A. and J. Pérez, (1979), “Within month predictions for monetary aggregates and Spanish monetary policy implementation”, 4th meeting of the Association d´Econometrie Appliquée, Rome, February.
- Espasa, A., (1980), “Las elasticidades de precio y renta en la determinación del efectivo en manos del público”, Estadística Española, No. 89, December.
- Espasa, A., J.R. Cancelo, (1987), “Un modelo diario para la predicción de la circulación fiduciaria”, unpublished, Servicio de Estudios, Bank of Spain.
- Espasa, A. and J. Salaverría, (1988), “Métodos cuantitativos para el análisis de la coyuntura monetaria en la economía española”, Boletín Económico, February.
A study evaluating salary inflation models up to 1977 can be found in Espasa (1977).
Espasa, Matea, Manzano and Catasús (1987) develop and generalise the approach proposed by Espasa, Molina and Ortega (1984), considering the need to analyse inflation for forecasting and diagnostic purposes, by means of a disaggregate econometric system. The paper was conclusive in a series of results that established the way that all subsequent analysts worked in relation to inflation in Spain. The paper provides a definition for core inflation that was later used by theBank of Spain, Spanish Instituto Nacional de Estadística, Eurostat and ECB. In this context, Espasa and Matea (1991) define the methodology for the calculation and forecasting of core inflation.
Espasa and Lorenzo (1995) progress in the disaggregate approach for forecasting inflation and in the introduction of leading indicators in specific price indices. These models are used in Espasa, Lorenzo and Escribano (1995) to defend the use of forecasts as a basis for diagnosing inflation, and, in Espasa and Lorenzo (1995) to analyse the convergence of Spanish inflation with Europe.
Espasa, Senra and Albacete (2002) approach inflation analysis and forecasting in the euro area, showing that models based on sector or geographic disaggregation generate more accurate forecasts than aggregate models. The paper also analyses the interest of constructing vector models for each type of disaggregation, including the long-term constraints affecting the different components. In the analysis of euro area inflation, Espasa and Albacete (2007) show that it is important to apply joint disaggregation by sector in each different country. A minimal sectoral and geographic disaggregation is analysed, so that it can be implemented with the available sample sizes while including the principal common trend factors among CPI components at their maximum level of disaggregation. This leads to a vector of ten components, two sectors in five geographical areas, constructing a VEqCM model with a block diagonality constraint. The paper shows that this model forecasts euro area inflation more precisely than any of the other aggregate or disaggregate alternatives considered. Albacete and Espasa (2005) approach the problem of forecasting euro area inflation as precisely as possible while explaining the factors which determine inflation forecasts. The paper shows that the most precise forecasts are obtained with monthly models of vector time series with an equilibrium correction mechanism, as proposed in Espasa and Albacete (2007). These models do not provide an explanation of inflation in terms of its causal variables. This is possible in Albacete and Espasa (2005) by means of quarterly vector econometric models with a mechanism to correct the equilibrium among the vector’s different economic variables. This paper proposed a way of combining the results of both types of model to obtain precise forecasts with a causal explanation.
Tena, Espasa and Pino (2010) approach the ambitious goal of forecasting the Spanish CPI, together with all its individual indices, with reference to all consumption sectors in each autonomous region. This and subsequent a but unpublished papers form the foundation for a thorough analysis of a region’s inflation and prices relative to the euro area, Spain and other similar regions. In turn, these relative prices are the basis for competition studies.
Mayo and Espasa (2011). The debate about forecasting an aggregate variable, directly or indirectly, is usually centred only on the forecasting accuracy of the aggregate. In contrast, the starting point of this paper is that all data matter -aggregate and components. The paper is focused on providing joint consistent forecasts for the aggregate and its components and in showing that the indirect forecast of the aggregate is at least as accurate as the direct one. The procedure developed in the paper is a disaggregated approach based on single-equation models for the components, which take into account common stable features which some components share between them. The procedure can be easily extended to include exogenous variables.
The procedure is applied to forecasting euro area, UK and US inflation and it is shown that its forecasts are significantly more accurate than the ones obtained by the direct forecast of the aggregate or by dynamic factor models.
A by-product of the procedure is the classification of a large number of components by restrictions shared between them, which could be also useful in other respects, as the application of dynamic factors, the definition of intermediate aggregates or the formulation of models with unobserved components.
The disaggregated inflation forecasts could show market differences, providing then, hopefully, some clues about the main factors causing inflation. But the procedure presented in this paper does not include causal economic variables to explain inflation, because in general the results from the economic theory will refer to the aggregate but not to the basic components. If forecasts for the aggregate are available from a causal model, they can be linked to the ones, usually more accurate, from the disaggregation approach by means of a regression, obtaining then the most accurate forecast with an econometric explanation.
Publications in international academic journals:
- Espasa, A., Senra, E. and Albacete, R. “Forecasting EMU inflation: A disaggregated approach by countries and by sectors”, The European Journal of Finance, v. 8, pages 402-421.
- Espasa,A. y R. Albacete (2007) ,"Econometric Modelling for Short-Term inflation Forecasting in the Euro Area ", Journal of Forecasting, v 26, pp 303-316, August. First WP version 2004.
- Tena,J.D., A. Espasa and G. Pino, 2010, “Forecasting Spanish Inflation Using the Maximum Disaggregation Level by Sectors and Geographical Areas”, International Regional Science Review, v 33, n 2, 181-204, April.
Other Publications and Working papers:
- Espasa, A., (1977b), “Modelos econométricos de inflación salarial”, Económicas y Empresariales, No. 3.
- Espasa, A., A. Molina and E. Ortega, (1984), “Forecasting the rate of inflation by means of the consumer price index”, Fourth International Symposium on Forecasting, London, July.
- Espasa, A., M.L. Matea and M.C. Manzano, (1987), “La inflación subyacente en la economía española: estimación y metodología”, Boletín Económico, March.
- Espasa, A. and M.L. Matea, (1991), “Underlying inflation in the Spanish economy: estimation and methodology”. Italian translation published in Note Economiche, v. XXI, No. 3, pages 477-493.
- Espasa, A. and F. Lorenzo, (1995), “Convergencia con Europa en la tasa de inflación: importancia, perspectivas y medidas económicas necesarias” Cuadernos de Información Económica, No. 100, July.
- Espasa, A., F. Lorenzo y C. Escribano, (1995), “La predicción de la inflación como base para un diagnóstico actualizado”, Revista del Instituto de Estudios Económicos, nº 1 y 2, pp. 331-366.
- Albacete, R., A. Espasa, (2005), ”Forecasting Inflation in the Euro Area using monthly time series models and quarterly y econometric models”, Working paper nº 2005-001. Departamento de Estadística. Universidad Carlos III de Madrid.
Mayo, I. and A. Espasa (2011), “Forecasting aggregate and disaggregates with common features”, working paper, Statistics Department, Carlos III University.
Espasa (1989) develops a procedure for estimating the trend of the Spanish gross domestic product, contemplating the existence of structural changes in mean growth, and Martinez and Espasa (1998) use that methodology to contemplate that the economic cycle is asymmetrical and requires modelling by non-linear structures. Espasa (1993a-b, 1994a-d and 1995a-d) includes A. Espasa’s contributions to “Economic Forecasts” about Spanish economy forecasting, and Espasa (1996) summarises his contribution, as the Macroeconomic Forecasting representative, to the group of Economic Forecasting experts created by the Ministry of Economy.
Minguez and Espasa (2006) analyse forecasting the euro area gross domestic product by different types of model, time series and econometric, with different disaggregations, showing that the best forecasting results are obtained by combining the GDP forecasts derived from disaggregation in terms of demand components with those derives from disaggregation in terms of supply components, using leading indicators whenever possible.
Cuevas et al. (2011) The paper proposes a methodology to estimate and forecast the GDP of the different regions of a country, providing quarterly profiles for the annual official observed data (RA). Thus the paper offers a new instrument for short-term monitoring that allows the analysts to quantify the degree of synchronicity among regional business cycles.
Technically, the paper combines time series models with benchmarking methods to forecast short-term quarterly indicators and to estimate quarterly regional GDPs ensuring their temporal consistency and transversal consistency with the National Accounts data (QNA). The methodology addresses the issue of non-additivity taking into account explicitly the transversal constraints imposed by the chain-linked volume indexes used by the National Accounts and provides an efficient combination of structural as well as short-term information.
The methodology is illustrated by an application to the Spanish economy, providing real-time quarterly GDP estimates and forecasts at the regional level. It is worth to emphasize that, from an operational perspective, timely forecasts of quarterly regional GDPs may be available with a minimum delay with respect to the national quarterly GDP release. In this way, the national figure may have appropriate regional counterparties, enhancing the informational content of analysis carried out at the aggregate level.
The main contributions of our paper are:
- A set of quarterly GDP figures at the regional level, derived in a consistent way with the official available data provided by the National Accounts, both RA and QNA.
- Early (or flash) estimates of quarterly GDP at the regional level that may be released at the same time as the national GDP.
- Short-term forecasts of quarterly GDP at the regional level by conditioning them on the projected path of the underlying short-term quarterly regional indicators.
Publications in international academic journals and books:
Espasa, A., (1989), “The estimation of trends with breaking points in their rate of growth: the case of the Spanish GDP”, published in Statistical Methods for Cyclical and Seasonal Analysis, edited by R. Mentz et al, 1989, Interamerican Statistical Institute.
Minguez,R. and A.Espasa (2006) “A Time Series Disaggregated Model to Forecast GDP in the Eurozone”, Chapter 17, pp 213-220, en Growth and Cycle in the Eurozone, Mazzi, G.L. and G. Savio (eds), Palgrave, December.
Martínez, J.M. and A. Espasa, 1998, “Caracterización del PIB español a partir de modelos univariantes no lineales”, Revista Española de Economía, v. 15, n. 3, pages 325-354.
Espasa, A. (1993a), “The outlook of the Spanish economy in the first quarter of 1993”, Economic Forecasts, July, pages 21-24.
Espasa, A., (1993b), “The Spanish economy: perspectives at the end of 1993”, Economic Forecasts, October, pages 24-25.
Espasa, A., (1994a), “Perspectives of the Spanish economy at the beginning of 1994”, Economic Forecasts, January, pages 20-21.
Espasa, A., (1994b), “Domestic and foreign demands in the Spanish economy for 1994”, Economic Forecasts, April.
Espasa, A., (1994c), “The state of an incipient recovery in the Spanish economy”, Economic Forecasts, July, pages 20-22.
Espasa, A., (1994d), “Spanish economy: good prospects, healing further domestic disequilibria”, Economic Forecasts, October, pages 23-25.
Espasa, A., 1995a, “The Spanish economy in 1995”, Economic Forecast, January 1995.
Espasa, A., 1995b, “The firm recovery of the Spanish economy waiting for more fiscal and structural reforms”, Economic Forecast, May 1995.
Espasa, A., 1995c, “Stable growth around three percent depending on the recovery of consumption”, Economic Forecast, August 1995.
Espasa, A., 1995d, “Stable growth around three percent in 1995 and 1996”, Economic Forecasts, November 1995.
Espasa, A., (1996), “Inflación y política económica”, in Proyecciones Macroeconómicas 1996-97, pages 19-29, Group of Economic Forecasting Experts, Ministry of Economy, 11 July.
Cuevas, A., E. M. Quilis and A. Espasa, 2011, “Combining Benchmarking and Chain-Linking for Short-Term Regional Forecasting”, paper accepted for presentation at the International Symposium on Forecasting, Praga, june.
Forecasting weekly, daily and hourly series of economic activity, for business logistics and planning, for the implementation of business intelligence and for the construction of real-time quantitative macroeconomic indicators
The first study of the forecasting of daily series of economic activity, Espasa (1997), was conducted in 1979 to evaluate the effects of the February banking sector strikes on the computable liability figures provided to the Bank of Spain. It was thus possible to estimate banks’ real computable liabilities and apply monetary policy accordingly, and not based on the declared liabilities that did not consider the enormous number of account movements that were not accounted for because of the strike. This subsequently gave rise to a thorough study on the modelling of monetary variables, which Espasa and Cancelo (1987) applied to the daily fiduciary currency circulation forecasting that the Bank of Spain needed to plan its cash transfers to banks.
Experience in the field of daily monetary series enabled an approach to daily electricity consumption forecasting in a 1988 project for Red Eléctrica de España. The model structure designed in this project is still used in REE and has forecast the historic peaks observed in the last few years. Unpublished studies conducted with J R Revuelta show that these forecasting procedures based on econometric models with complex structures in order to consider calendar, holiday, seasonal and weather-related effects, generate much more accurate forecasts than those obtained from methods based on neuronal networks. The theoretical development required in the econometric models used in the 1988 project with regards to a weather variable effect that is non-linear and dynamic and changes with the type of day and season of the year, is described in Cancelo and Espasa (1995). Cancelo et al. (2008) present the models for hourly electricity consumption forecasting and design a strategy for combining the results of a daily model and 24 hourly models for forecasting consumption in the 168 hours of the next seven days. Taylor and Espasa (2008) provide a general description of the quantity and price forecasting problems currently affecting the electric sector, including those associated to renewable energy, the production of which cannot be controlled.
Espasa et al. (1996) and Cancelo and Espasa (1996) generalise the models developed for daily monetary and electricity consumption series for many other series. Cancelo and Espasa (2001) study high-frequency (daily, hourly, etc.) time series containing valuable information for business forecasting and logistics. The paper proposes and illustrates the use of time series techniques for that purpose, developing a general framework capable of reflecting the principal properties of these time series. These techniques are being widely used by major international corporations in sectors such as energy, transport and communications in order to forecast different variables that are of importance for their respective businesses. A field in which such forecasting is very useful is satisfying the demand of major clients that need to be supplied on a weekly or daily basis. Indeed, based on the historic figures relative to their customers, suppliers are able to forecast their weekly or daily needs, ensuring sufficient supply.
In line with the above, Cancelo and Espasa (2009) develop thee basic ideas for the implementation of Business Intelligence techniques, showing how to develop a forecasting system that converts a firm’s raw data in useful knowledge for market operations.
The possibility of having real-time daily information about economic activity variables such as electricity consumption, consumption of other sources of energy, communications, transport, transactions, sales, etc., means that it is possible to construct hard macroeconomic indicators that quantify the magnitude in question in real time. These indicators are very important, as they can be obtained as soon as soft indicators that merely synthesise a majority opinion concerning their evolution. The construction of this hard data, however, is much more complex than merely compiling daily observations, as the magnitudes to which they refer are highly affected by weather and the calendar, and these effects are not macroeconomic indicators. It is therefore essential to correct daily observations for such effects and construct indicators from the corrected data. Cancelo and Espasa (2009) present a method based on daily and monthly models enabling the construction and forecasting of such hard data. The method also enables advancing the indicator for the month in question much before it ends.
References in international academic journals and books:
Espasa, A., J.M. Revuelta and J.R. Cancelo (1996), “Automatic modelling of daily series of economic activity”, in Prat, A. (ed) Proceedings in Computational Statistics, pages 51-64, Physica Verlag, Heidelberg.
Cancelo, J.R. and Espasa, A. (1996) “Modelling and forecasting daily series of economic activity” Investigaciones Económicas, V. XX(3), pages 359-376.
Cancelo J R, A Espasa and R Grafe, 2008, Forecasting the electricity load from one day to one week ahead for the Spanish system operator, International Journal of Forecasting, v 24,n 4, pages 588-602.
Taylor, J W and A. Espasa, 2008, “Energy Forecasting”, introduction to the special issue on Energy Forecasting, International Journal of Forecasting, v 24, n4, pages 561-65.
Cancelo,J R and A Espasa (2009), “Implementing Business Intelligence in Electricity Markets”, in Wang S and J Wang (eds.) Business Intelligence in Economic Forecasting: Technologies and Techniques, IGI Global.
Espasa, A., (1979b), “Un modelo diario para la serie de depósitos en la Banca: primeros resultados y estimación de los efectos de las huelgas de febrero de 1979”, Boletín Económico, Bank of Spain, July-August, 1979, pages 33-37.
Espasa, A., J.R. Cancelo, (1987), “Un modelo diario para la predicción de la circulación fiduciaria”, unpublished, Servicio de Estudios, Bank of Spain.
Cancelo, J.R. and A. Espasa, (1995), “Modelización del efecto temperatura en el consumo de electricidad: un ejercicio de búsqueda de especificación en relaciones dinámicas no lineales”, Revista Estadística Española, v. 37, No. 139.
Cancelo, J.R. and Espasa, (2001) A., ”Using high-frequency data and time series models to Improve yield management “, International Journal of Services Technology and Management, v. 2, pages 59-70.
Cancelo, J C and A. Espasa (2009), “Modelling Monthly Electricity Demand: Interactions Between Monthly Econometric and daily Time Series Models”, presented at the 29th International Symposium on Forecasting, Hong-Kong, June.
Macroeconomic forecasting methodology.
The outstanding contribution of the book by Espasa and Cancelo (eds) (1993) consists of its in-depth presentation of all the available statistical-econometric methods and the creation of a methodology for its application to analysing economic outlook. The previous papers that influenced this book included, among others, Espasa (1977a- c, 1978, 1980a, 1980b, 1982a and b and 1983). In the book’s prologue, E. Fuentes-Quintana expresses an opinion that the subjects contemplated in the book should be subject to examination at the end of B.Sc. courses in economics. A subsequent contribution concerning forecasting and macroeconomic analysis methods can be found in Espasa and Albacete (2004). Espasa and Peña (1995) present a procedure for breaking down the forecasting function of an ARIMA model into one permanent and one temporary term, comparing it with other methods presented in the literature and applying it to actual series. The paper generalises the concept of integration with the I(d,m) notation for distinguishing the stochastic and deterministic factors in the permanent component of an economic variable.
References in international academic journals and books:
- Espasa, A., (1977), “A note on the acceptability of regression solutions: another look at computational accuracy”, Journal of the American Statistical Association, v. 72, No. 359, page 603, September.
- Jenkins Analysis” in “ An Eponymous Dictionary of Economics :a Guide to Laws and Theorems Named after Economists ”,Segura,J. y Rodriguez Braun C. (editors).
- Espasa, A., (1982), “Relationships between variables: the short and long run effects”, chapter 13 in Selected Papers on Contemporary Econometric Problems, edited by E.G. Charatsis, The Athens School of Economics and Business Science.
- Espasa, A. and Peña, D., 1995, “The decomposition of forecast in seasonal ARIMA models”, Journal of forecasting, 1995, December, pages 565-584.
- Espasa, A. and J.R. Cancelo, (1993), Métodos Cuantitativos para el Análisis de la Coyuntura Económica, Alianza Editorial, Madrid.
- .Espasa, A., (1977b), “Modelos econométricos de inflación salarial”, Económicas y Empresariales, No. 3.
- Espasa, A., (1977c), “El problema de la desestacionalización de las series económicas. Métodos utilizados y su interpretación”. Boletín de Estudios Económicos, v. 32, nº 101, pags. 471-478 agosto.
- Espasa, A., (1978a), Estimación y selección de modelos econométricos dinámicos, Bank of Spain. Servicio de Estudios. Estudios Económicos, No. 11.
- Espasa, A., (1978b), El paro registrado no agrícola 1964-1976: Un ejercicio de análisis estadístico univariante de series económicas, Bank of Spain. Servicio de Estudios. Estudios Económicos, No. 15.
- Espasa, A., (1980a), La predicción económica, Bank of Spain. Servicio de Estudios. Estudios Económicos, No. 18.
- Espasa, A., (1980b), “Las elasticidades de precio y renta en la determinación del efectivo en manos del público”, Estadística Española, No. 89, December.
- Espasa, A., (1982b), “Un estudio econométrico de la tasa de variación del empleo en la economía española”, in El mercado de Trabajo en España, Secretaría General Técnica, Ministry of Economy and Trade, pages 271-324.
- Espasa, A., (1983), Un estudio econométrico de la tasa de variación del empleo en la Economía Española, Estudios Económicos, Bank of Spain, Madrid.
- Espasa, A. and Albacete, R., 2004, “Consideraciones sobre la predicción económica: metodología desarrollada en el Boletín de Inflación y Análisis macroeconómica”, pages 635-660, Políticas, Mercados e Instituciones Económicas, Pérez, J., Sebastián, C. and Tedole , P. (eds), Editorial Complutense, Madrid.