Estimating the size of the shadow economy in Spain: a structural model with latent variables

Abstract

There has recently been a revival of international interest in measuring the size of the shadow economy. The current study adopts an approach to the Spanish case that is based on the theory of unobservable variables. This methodology involves the estimation of structural models (MIMIC) which analyses a set of causes of the shadow economy while simultaneously taking into account its influence upon a series of indicators. The proposed model permits the determination of a relative evolution over time of the size of the shadow economy, which requires the calibration of the model with an exogenous estimation in order to obtain real values. The exogenous estimation employed is that obtained by a monetary method based on a money demand function. The results show a considerable shadow economy, measuring between 8 and 18.8% of GDP in the period 1976–2002, and demonstrate that the shadow economy is significantly influenced by the tax burden, the degree of regulation and unit labour costs. A positive correlation is obtained between GDP, money demand and the level of the shadow economy.

Acknowledgements

An earlier version of this paper was published as documento de trabajo de la Fundación de las Cajas de Ahorros (FUNCAS) 184/2004 (Working paper of Savings Bank Foundation, FUNCAS, 184/2004) according to their Programa de Estimulo a la Investigación (Pro-research Program). The authors wish to thank the help of Roberto Dell’Anno, the useful commentaries of two anonymous referees from FUNCAS and from Applied Economics and the financial support of Instituto de Estudios Fiscales.

Notes

For a more detailed and exhaustive classification, see Schneider and Enste (2000).

This is why these models are also known as ‘covariance structure analyses’.

The details may be consulted in Appendix E of Dell’Anno (2003).

Bhattacharyya (1990) presented a very comprehensive study of estimation methods based on money demand.

A more detailed analysis of the origin of these results is presented in Alañón and Gómez-Antonio (2003).

In Frey and Weck-Hannemann, 1984; Giles, 1995; 1999a; or Dell’Anno, 2003, for example.

The OECD Statistical Compendium only has Spanish fiscal data from 1976 onwards.

In the majority of structural models, these effects may be decomposed into direct and indirect effects. In this model, the decomposition is not called for, since the existence of only one latent variable means there are no indirect effects, and the total effects thus coincide with the direct effects.

In order for the economic interpretation of the resulting rates to be more direct, we have reestimated the model without standardizing the variables, subsequently applying the coefficients to the unstandardized variables.

When the residuals are high, the program identifies the variables responsible for the covariance residuals being higher.