Endogenous exchange rates in empirical stock-flow consistent models for peripheral economies: an illustration from the case of Argentina

Abstract

In recent years there has been an increase in empirical SFC models applied to specific countries to address a broad range of research questions. However, in most cases, the exchange rate has been modeled as exogenous. This prevents the analysis of both the multiplicity of channels through which the global financial cycle impacts economies and the effects that the different tools of domestic economic policy have on the external sector and macro-financial stability. To overcome this limitation, this paper presents an empirical quarterly SFC model for Argentina where the exchange rate is modeled among the lines proposed by Godley and Lavoie. The model includes a diversity of tools of economic policy that in the recent past have affected exchange rate stability and, consequently, the economy in general. The results suggest that the closure used in this model can be useful for the construction of similar models for other peripheral countries in which the nominal exchange rate results from the interaction of different processes from both the domestic economy and the rest of the world.

Keywords:

• Argentina
• economic policy
• exchange rate
• stock-flow consistent models

JEL CLASSIFICATION CODES:

• C54
• E17
• E61

Notes

1 Although this tax is paid by households when they consume final goods and services, their inclusion in this section of the matrix follows from the need to be consistent in the valuation of the different variables. While the components of aggregate demand are valued at market prices, the variables included in the income generation account are written at basic prices. Moreover, the deduction of the value-added tax from the value of the final sales is necessary because otherwise the gross operating surplus would be overestimated.

2 For the sake of simplicity it is assumed that the cost of production and the electricity subsidy rate and that of the other utilities are the same. Future versions of the model should provide a more detailed description of the cost structure of the different utilities where the government has a price regulation policy. Yet, splitting utilities into electricity and others in the current version of the model allows for more precise econometric estimations of the demand equations.

3 Mixed income is taken as part of this income source given the large number of professional workers that undertake their activity in the form of self-employment.

4 Although theoretical SFC models work with Tobinesque portfolio equations of the type proposed by Brainard and Tobin (1968) and Tobin (1969), it is not easy to build empirical approximations to the rate of return matrix embedded in this approach and to obtain estimators that respect its constraints. Some researchers, like Zezza and Zezza (2020), attempt to adapt the structure of the model to keep the main concepts embedded in Tobin’s theory. In this model a simpler path is taken by estimating equations separately with the caution of obtaining estimations that are consistent with the intuitions underlying those decisions.

5 These equations are estimated using the nonlinear autoregressive distributed lags methodology, as proposed by Shin, Yu, and Greenwood-Nimmo (2014). This methodology assumes that the reaction of the dependent variable to changes in the independent variable varies according to the direction in which the change goes. For instance, the variable in the third term of the right-hand side of Equation (46)(46) $$\Delta \mathit{log}\left(\frac{{DD}_t}{W_{t-1}^P}\right)={\gamma }_0+{\gamma }_1\hspace{0.17em}\mathit{log}\hspace{0.17em}\left(p_{t-1}^C\right)+{\gamma }_2\hspace{0.17em}\mathit{log}\hspace{0.17em}\left(r_{t-1}^{G+}\right)+{\gamma }_3\hspace{0.17em}\mathit{log}\hspace{0.17em}\left(r_{t-1}^{G-}\right)+{\gamma }_4{\mathit{controls}}_{t-1}++{\gamma }_5\Delta \mathit{log log}\mathrm{ }\left(\frac{{DD}_{t-1}}{W_{t-2}^P}\right)\mathrm{ }+{\gamma }_6\Delta \mathit{log}\left(p_t^C\right)+{\gamma }_7\Delta \mathit{log}\left(r_t^{-}\right)+{\gamma }_8\Delta \mathit{log}\left({\mathit{vol}}_{t-1}^E\right)$$(46) gathers the accumulated sum of all the positive changes in the interest rate on bonds. On the other hand, in the variable in the fourth term, where the “−” sign appears, the sum of all the negative changes in the interest rate on bonds is computed.

6 The component that explains the possibility of the central bank having a positive profit is the capital gains that arise from its holdings of foreign reserves (as shown in the revaluation matrix presented in section 4). When the exchange rate depreciates, the central bank can monetize these capital gains and transfer the resulting funds to the government.

7 The model’s capacity to track the actual trajectory of a variable can be tested for any of the endogenous variables of the model. For the sake of simplicity only the model validation for GDP and private saving are presented. These two variables are among the most endogenous variables of the model as they are entirely composed of other endogenous variables, some of the behavioral and some of them accounting identities.

8 Thiel's inequality coefficient provides a measure of how well a time series of estimated values compares to a corresponding time series of observed values. It ranges between 0 and 1, with 0 representing a perfect fit.

9 Capital controls are defined as a dummy variable that takes the value of 1 between 2012Q1 and 2015Q4. This variable is included in the long-term equations of the exchange rate, which residual is explicitly written in the short-term relation as reflected in Equation (59)(59) $$e_t=\frac{E_t{p}_t^{\mathit{USA}}}{p_t^C}$$(59) .

10 The positive relationship between the real exchange rate and exports needs not be explained through an increase in the sales of manufactured goods to foreign markets, but by the liquidation of primary exports that are often accumulated as a stock to be sold only once a devaluation has occurred, thereby making a capital gain.