Forecasting regional GDPs: a comparison with spatial dynamic panel data models

ABSTRACT

The monitoring of the regional (provincial) economic situation is of particular importance due to the high level of heterogeneity and interdependences among different territories. Although econometric models allow for spatial and serial correlation of various kinds, the limited availability of territorial data restricts the set of relevant predictors at a more disaggregated level, especially for gross domestic product (GDP). Combining data from different sources at NUTS-3 level, this paper evaluates the predictive performance of a spatial dynamic panel data model with individual fixed effects and some relevant exogenous regressors, by using data on total gross value added (GVA) for 103 Italian provinces over the period 2000–2016. A comparison with nested panel sub-specifications as well as pure temporal autoregressive specifications has also been included. The main finding is that the spatial dynamic specification increases forecast accuracy more than its competitors throughout the out-of-sample, recognising an important role played by both space and time. However, when temporal cointegration is detected, the random-walk specification is still to be preferred in some cases even in the presence of short panels.

KEYWORDS:

• Prediction
• spatial correlation
• panel data
• regional GVA forecasting

JEL classification:

• C21
• C23
• C53

ACKNOWLEDGEMENTS

This paper began as Tomelleri’s doctoral thesis chapter. We thank three anonymous reviewers and the associate editor for their careful and insightful comments on our paper and, indeed, on the development of the whole work. We also thank all the participants at the European Regional Science Association (ERSA) Conference for their comments. All remaining errors are ours.

DATA AVAILABILITY

Data and codes are available at http://francescoravazzolo.com/pages/research.html

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

Notes

1 New provinces become effective in some Italian regions between 2004 and 2009: the province of Barletta-Andria-Trani in 2004, four new provinces in Sardinia in 2005 and Fermo e Monza-Brianza 2009.

2 It is not possible to obtain real GVA at the provincial level. We obtain the deflator as the ratio of nominal and real GVA at the regional level and then assign it to the provinces within each region. The results do not change estimated coefficients and the forecasts that much, and are available upon request.

3 This is in line with Eurostat (Structural business statistics (SBS) size class).

4 International Tourism Highlights, 2019 Edition.

5 More precisely, GVA is defined by Eurostat as the value of output less the value of intermediate consumption.

6 For a broader discussion on the different weights specifications of the matrix W, see also Giacomini and Granger (2004)

7 We also tried an autoregressive (AR) specification. We found that our model outperforms both specifications and there are no substantial differences between AR and RW models. Then, we decided to drop the AR specification from the entire analysis. Results are available upon request.

8 The size classes of firms with more than 49 employees has changed several times between 2000 and 2016; see section 3.