Regional output growth and the impact of macroeconomic shocks in Mexico

Abstract

This paper presents an empirical analysis of the relationship between national and regional output growth in Mexico, and the impact of domestic and international shocks on national, regional and state output movements. Our results suggest that there are similarities, but also significant differences, in real output dynamics across the regions and states of Mexico and that it would be wrong to regard the Mexican economy as a homogeneous entity. The results show that real output growth in Mexico and the United States are linked, but there is no common output trend for the two countries. At the regional level, it appears that North and Central Mexico share similar features, but the path of output growth is more distinctive in South Mexico. Overall, our results suggest that assessments of macroeconomic performance, and related discussions of policy, should pay greater attention to the potential diversity in regional performance.

Keywords:

• regional output
• macroeconomic shocks
• trends and cycles
• co-movement
• dynamics

JEL Classifications:

• F41
• R11
• C22
• E32

Acknowledgements

We would like to thank the editor and the two anonymous referees for useful comments and discussions. Any remaining errors are ours.

Notes

1. Representative examples of work in these areas include: Martin and Sunley (1998); Martin (2001); Shepherd and Dixon (2002); Barrios and de Lucio (2003); Rodriguez-Fuentes and Dow (2003); Carlino and DeFina (2004); Gardiner, Martin, and Tyler (2005); Pons-Novell and Tirado-Fabregat (2006); Suedekum (2006); Dow and Montagnolia (2007); Montoya and De Haan (2008); Owyang, Rapach, and Wall (2009); Wilkerson (2009).

2. A useful review of the various shocks affecting the Mexican economy through the 1980s and the 1990s can be found in Gould (1995).

3. The series for Mexican real GDP per state were generated by combining the series published by the National Institute of Statistics, Geography and Informatics (INEGI) of Mexico for the 1993–2006 period (base year = 1993) with a methodology of projection (see Mendoza 1997, 2013) that used census data by sector level of economic activity for the years 1970, 1975, 1980, 1985, 1988 and 1993 at current prices. This methodology considers the estimated structural and spatial distribution functions of GDP by state for each of the 73 sectors of economic activity (plus imputed bank services) and applies them to the National Accounts for the period 1970–2006. Regional data are available only on an annual basis and so we are not able to conduct the analysis on a quarterly basis. It is also important to mention that our sample period cannot be extended beyond 2006 because of a change in the way that INEGI compile the data after this date. Significant changes include different industrial classification systems, the addition of services to the economic activities and a new accounting scheme of oil production. The series for US real GDP was obtained from the Federal Reserve Bank of St Louis data bank. The computations reported in the paper were undertaken in Matlab 5.2 and Eviews 6.

4. It has been argued for example that the states geographically close to the US (border states) and the states in the Central region with competitive industries arising from economies of agglomeration have achieved higher growth than the regional average and that the gap between these states and others in the regions has widened (Esquivel 1999; Diaz-Bautista 2003).

5. This statement is generally true, but there are perhaps three special cases. The first two cases are Campeche and Tabasco, where the economy is based mainly on the extraction of oil and its performance dynamics is presumably determined by conditions in the international market, which are linked to US demand. The third case is Quintana Roo, where economic activity is heavily dependent on international tourism (focused mainly in Cancun and Playa del Carmen), which originates mainly from the US and Canada, and also Europe.

6. The absence of cointegration between Mexico and the United States points to the absence of cointegration between the regions and the US, but the test result for the national series could in part reflect the differing trend structures of the component regions.

7. Campeche is a state for which oil production is important and in Quintana Roo tourism is significant.

8. Alternative tests for cointegration have been also proposed by Larsson, Lyhagen, and Läothgren (2001) and McCoskey and Kao (1998). The former test is based on Johansen’s (1998) maximum likelihood estimator whereas the latter uses a Lagrange multiplier test on the residuals.

9. Some studies have used the methodology of panel cointegration to test opening and convergence hypotheses for the same period of analysis (1985–2006). These studies point to changes in the pattern of regional dynamics and convergence over the period. See Gómez and Ventosa (2009) and Díaz, Sánchez, and Mendoza (2009).

10. The weak instrumentation problem is discussed in Han, Phillips, and Sul (2010). They also propose a solution to the problem utilizing a variable differencing procedure they refer to as ‘X- Differencing’.