Regional growth under a monetary perspective: a theoretical model with empirical application to the Brazilian case

Abstract

This study aims to establish a link between the regional growth rate and the Keynesian concept of Liquidity Preference from a model of Circular Cumulative Causation. Particularly, it replaces Verdoorn's Law by a mutual causal relationship between credit and regional growth rate. The model is represented by a first-order difference equation dynamic system. The analytical solution suggests that regional changes in Liquidity Preference lead to a Center/Periphery pattern of growth. An empirical application is performed for the Brazilian case, using banking variables at state level in a GMM estimation for Dynamic Panel Data. The results corroborate the hypothesis of persistent different growth rates between the more and the less-developed Brazilian regions.

Keywords:

• Circular cumulative causation
• liquidity preference
• regional growth

Notes

Notes

1 It is worthwhile to make an additional comment regarding the signal of ε_i. It can eventually assume positive values depending on the phase of the cycle as well as on the type of the relationship between the regions i and j. For instance, if in equation 2 i is the periphery and the economy is growing, the central region's spread effects over the peripherals implies a positive state of confidence for the whole economy, resulting in ε_i >0. In other words, a rise in the expected return on investment in the centre positively affects the availability of finance in the periphery. One evidence of this behavior are the findings of Rodriguez-Fuentes and Dow (2003) which indicated that, in periods of expansion/recession, credit tends to increase/decrease more than proportionally in the periphery.

2 The interest rate adopted in this model follows the properties of instant capitalization at variable compound interest. Unlike periodic capitalization, in which interest due is credited only at the end of each period, in instant capitalization interest is credited at each infinitesimal interval of time and, as a result, the amount continually increases over time. Moreover, in this capitalization model, it is assumed that interest rates may vary over time. Using this interest function assigns generality to the model, allowing it to fit any region, regardless of the evolutionary characteristic of the interest on time - fixed or variable. Thus, under the hypothesis of instant capitalization, a monetary unit invested in $\tau =0$ will result in ($1,00 ) $e^{\left({\sum }_{\tau =0}^t{s}_{\tau }\Delta s\right)}$ in t periods. In its continuous form it is given by $S_t=\left({\int }_{\tau =0}^t{s}_{\tau }d\tau \right).$ Further details on instant capitalization can be found in Bueno, Rangel and Santos (2011).

3 Rodriguez-Fuentes and Dow (2003) analyze the European monetary unification and its effects on the credit market stability in different regions of Spain.

4 The general solution has the form $y_t=A{b}^t+c$ where $b=\mathit{\text{γδλA}}=y_0+\left[C/\left(\gamma \delta \lambda -1\right)\right]$ and c equals the second term in equation (11); notice that the latter is the equilibrium growth rate given by equation (8).

5 The delimitation of the period of analysis is justified by the availability of data by the Central Bank of Brazil, which begins in 1995, as well as by the country’s political stability in the period, considering that the popular manifestations that began in 2013 could influence the variables, making them not reflect their true paths.

6 Bolsa Família is a federal government transfer mechanism which aims to combat poverty by ensuring a minimal income under the commitment that families keep their children in school.

7 The Arellano and Bond (1991) approach allows for the use the lagged value of the explanatory variables as an instrument to control the endogeneity problem (caused by reverse causality), as well as, controls for unobserved fixed effects with a difference equation.

8 Notice that this term is the inverse of what is presented in the theoretical model. This change does not alter the meaning of the model since the elasticities remain the same. What changes is only the signal, making easer the interpretation once it allows for a direct association with the concept of liquidity preference (LP). Specifically, as discussed in section 3, an increase in non-banking agents’ LP raises DV_i relatively to DP_i.

9 According to the Arelano and Bond test, the null hypothesis of non-autocorrelation of the residuals for the upper lags is not rejected, indicating the consistency of the estimated parameters. In addition, the Sargan test reinforces the good fit of the model, indicating the non-rejection of the null hypothesis, which indicates that all the instruments used are significant to eliminate the endogeneity problem generated by the lagged dependent variable.

10 Appendix 3 presents the results for the autocorrelation (Abond) and instrument validity (Sargan) tests for the Restrict Model, which indicated the consistency of the adjustment.

Additional information

Notes on contributors

Teófilo de Paula

Teófilo de Paula is at Federal Rural University of Rio de Janeiro – UFRRJ, Três Rios, Brazil.

Fábio Gama

Fábio Gama is at Federal University of Juiz de For a, Juiz de Fora, Brazil.

Marco Crocco

Marco Crocco is at Federal University of Minas Gerais, Belo Horizonte, Brazil.