Level of services, spatial dependence and allocative efficiency in local governments

ABSTRACT

In the public economics literature expenditure needs, allocative efficiency and spatial dependence of local governments costs have been widely analysed separately implying bias estimations of the expenditure needs at local level. An original procedure that simultaneously takes into account the standard level of services, the allocative efficiency and the spatial proximity among Municipalities, has been proposed. The estimation strategy has been applied on a very detailed database of more than 4,000 Italian Municipalities for the year 2013.

KEYWORDS:

• Local governments
• fiscal federalism
• expenditure needs
• allocative efficiency
• duality
• Stochastic Frontier Analysis
• spatial autocorrelation model

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. ‘Decentralization holds a lot of promise, but whether it improves public service delivery depends on the institutional arrangements governing its implementation. Several conditions must be met before the full benefits of decentralization can be reaped. First, for decentralization to increase allocative and productive efficiency, local governments need to have the authority to respond to local demand as well as adequate mechanisms for accountability’ (Kahkonen and Lanyi 2001).

2. Please see Worthington and Dollery (2000) for a complete survey of frontier efficiency measurement techniques in local public sector.

3. Although in practice the level of service were not taken into account in the spending cuts criteria.

4. Requirements for each authority are determined by the linear combination of goods/services with a plurality of weights (prices); usually the choice of these factors is entrusted to experts or is submitted to a political decision.

5. For example, the external factors that, ceteris paribus, can favor or hinder the supply of local public goods, such as the morphological characteristics of the territory or the surface area.

6. Please note that the notation is different from the standard one to be consistent with the following paragraphs.

7. This approach is suitable for policies which aim to improve the overall performance of the LAs as a whole by progressively increasing the standard level of services given the available macro budget.

8. In the Section 6 we will refer, without loss of generality, to an univariate output .

9. The model is specular in the case of optimal capital price given the labor one.

10. The properties can be summarized as:

(i) and ;

(ii) is a upper semi-continuous function;

(iii) for ;

(iv) for ;

(v) is homogeneous of degree 1 in ;

(vi) if and only if ;

(vii) if belongs to the ‘frontier’ of the production possibility set.

11. For the sake of simplicity, refers to all covariates in the model.

12. is the spatial weight matrix summarizing neighborhood relations among the territorial units.

13. In literature this estimator is also named as: ‘Mixed-regressive-spatial autoregressive model with a spatial autoregressive disturbance, (SARAR)’.

14. More information at: http://tinyurl.com/cz9stat.

15. The ‘essential’ functions for the Italian Municipalities are: General Administration, Local Police, Education (complementary services), Public Roads and Transport, Planning and Environment and Social care.

16. In public services, in fact, we are often in the presence of multi-input and multi-output processes; in this case the different quality of the output can heavily affect the estimation of efficiency.

17. It tests the heteroscedasticity of the residuals, namely if the estimated variance of the residuals are dependent on the values of the independent variables.

18. It is a goodness-of-fit test, testing the normality of the residuals through the skewness and kurtosis.

19. Estimation based on the cost of renting properties at the provincial level.

20. Data source: IX General census of industry and services – Survey public institutions, ISTAT. The elementary data, available only for the total of the municipality, has been made proportional by the percentage of the costs of the Municipal Registry function on the total expenses for employees.

21. In the error components frontier formulation, see Battese and Coelli (1992).

22. In the estimation dependent variables and covariates are logged and divided by the mean.

23. The indirect impacts of on (that exerts on its neighbours , which in turn feeds back into ) are not yet available for SAC models in R spdep package.

24. Please note that we used the Kelejian and Prucha notation.

25. The authors thanks Roger Bivand and Martin Gubri for their advice regarding the predictions in spatial autoregressive models; more in particular we used the sppred function still not public in R spdep package – with the KP3 (Kelejian and Prucha 2007) predictor to obtain the best linear unbiased prediction (BLUP).

Additional information

Notes on contributors

Francesco Vidoli

Francesco Vidoli is an integrative course teacher in economic and statistics at the Faculty of Political Science, University of Roma Tre, Italy. His main research interests are spatial production efficiency and composite indicators nonparametric methods.

Elisa Fusco

Elisa Fusco is a subject expert in economic statistics for the course ‘Efficiency and productivity analysis’ at the University of Rome La Sapienza. Her research interests include efficiency and productivity analysis, spatial econometrics and methods of composite indicators construction through frontier models.