The Asymmetric Effects of Argentina’s Fiscal Deficits on the Real Exchange Rate

ABSTRACT

According to standard theoretical frameworks such as Real Business Cycle (RBC’s) models or new-Keynesian theories, the real exchange rate should appreciate in response to an increase in government spending. However, the empirical literature finds mixed results. We offer an answer to this puzzle by analyzing the impact of the composition of the fiscal deficit. Using a dynamic stochastic general equilibrium model with a government and an external sector, we quantify the differential impact on the real exchange rate generated by an increase in public consumption expenditure, public investment, and tax reduction. We calibrate and simulate the model for Argentina and find that the fiscal deficit originated in tax reduction can improve the real exchange rate. In contrast, one generated by an increase in spending deteriorates the real exchange rate. In particular, this depreciation is more significant when the spending is directed toward public consumption than when used for public investment. We argue that quantifying these different effects on the exchange rate within a dynamic stochastic general equilibrium framework is an essential exercise of political economy for highly dollarized emerging economies that exhibit higher inflation pass-through.

KEYWORDS:

• General equilibrium
• financial frictions
• public investment
• government spending
• tax reduction
• real exchange rate

JEL:

• E32
• E61
• H62

Acknowledgments

We appreciate helpful comments and suggestions from Ricardo Lagos, Robert King, and participants at the Politics and Economics of International Finance (PEIF) conference at Harvard Kennedy School, participants at the Development and International Economics Seminar at Boston University, and participants from the Society for the Study of Emerging Markets – ASSA Chicago 2021. Julio gratefully acknowledges financial support from Metropolitan College of Boston University.

Disclosure Statement

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

Notes

1. Bello, Heresi, and Pineda Salazar (2010) estimate the real exchange rate for 17 Latin American countries from 1970 to 2005 and demonstrate the existence of recurrent processes of exchange rate overvaluation for several of these countries.

2. The Balassa-Samuelson model (BS) was the first model that related sectoral productivity shocks to movements in the real exchange rate. In the BS framework, productivity in the tradable sector, given factor price equalization, determines the price of non-tradable goods. Baldi and Mulder (2004) confirm these predictions for Argentina, Chile, Brazil and Mexico.

3. Also, Chari, Kehoe, and McGrattan (2002) use a general-equilibrium model with sticky prices that can generate real exchange rates that are appropriately volatile and quite persistent, though not quite persistent enough. In their model, exchange rate fluctuations are generated by monetary shocks interacting with sticky-price goods. However, prices need to be held fixed for at least one year, and assuming that firms cannot change their prices for so long is in contrast with the new evidence on micro data (See Mackowiak and Smets 2009).

4. Bergoeing et al. (2002) use an open economy real business cycle model to describe the behavior of Mexico and Chile during the so-called “lost decade.” They rule out a monetarist explanation, an explanation based on falls in real wages and real exchange rates, and a debt overhang explanation. They conclude that the crucial difference between the two countries was the earlier policy reforms in Chile in banking and bankruptcy procedures that generated faster productivity growth.

5. Interestingly, J. Taylor (2019) shows that while interest rate rules work well to keep inflation low in a low-inflation regime, getting inflation down from high levels to low levels is a transition issue that raises other concerns. Taylor observes that it is very difficult to assess the level of the real interest rate when inflation expectations are moving around rapidly. Thus, he suggests that instruments such as the money supply or the monetary base are critical for such transitions citing as an example the recent change in the monetary policy rule at the Central Bank of the Republic of Argentina – from an interest rate rule to a monetary base rule.

6. To understand this, notice that policy deviations at the “center country” central banks in the form of “lower-than-rule” interest rates cause capital outflows from the “center” countries and thus inflows to emerging market countries along with appreciations of their currencies. Simulations by Carabenciov et al. (2013) show that the exchange rate appreciation effect dominates other effects and has a negative effect on emerging market countries. Thus, the emerging market central banks try to counteract this effect by lowering their own policy interest rates below their own policy rule (See Hofmann and Bogdanova 2012).

7. Regarding notation, observe that most of the variables come with a superscript. For example, in the variable $p^x$, we use the superscript ($x$) on the variable price $p$ to denote the price of exportable goods.

8. If ratios were different, then we would need to include a welfare function for the government to see how it allocates resources between tradable and nontradable goods to their various uses. If we had access to more granular data about the composition of public consumption or public investment, it would pay off to have a more complex model in which the consumer chooses between tradable and nontradable final goods via a CES function. For example, we could speculate that some types of public investment may have a greater impact on the traded sector, while others may largely affect the nontraded sector. This could have a differential impact on the real exchange rate worth exploring. However, the existing data on the composition of public investment is still limited, making harder the task of evaluating this more involved model (see Kappeler and Välilä 2008).

9. The amount of government debt held by domestic residents is equal to the amount of debt sold by the government to this group. Thus, market-clearing condition guarantees $B_t^{h,g}+B_t^{g,h}=0$, or equivalently, $B_t^{h,g}=-B_t^{g,h}$. Then, it is easy to see that the total debt for the country is $B_t=B_t^h+B_t^{\mathrm{g}}=\left(B_t^{h,x}+B_t^{h,g}\right)+\left(B_t^{g,x}-B_t^{g,h}\right)=B_t^{h,x}+B_t^{g,x}$. This last expression shows that the total debt is exactly the total (net) external debt. Notice that $B_t^{h,g}>0(<0)$ means that the family is borrowing (lending) funds.

10. Note that functional form in equation (6) is exactly equal to the uncovered interest rate parity (UIP) condition when the level of debt is equal to its steady-state level, i.e., $B_t=\bar{B}$; making the second term in (6) equal to zero as in García-Cicco, Pancrazi, and Uribe (2010).

11. Notice that equation (8) is mathematically equivalent to a model with a constant international interest rate and a shock to the country’s risk premium instead (See García-Cicco, Pancrazi, and Uribe 2010).

12. We derive in detail all the optimality conditions in the technical Appendix B.

13. Dawkins, Srinivasan, and Whalley (2001) argue that calibration is estimation and estimation is calibration. In page 3663, they write: “If calibration is the setting of the numerical values of model parameters relative to the criterion of an ability to replicate a base case data set as a model solution, and estimation is the use of a goodness of fit criterion in the selection of numerical values of model parameters, the two procedures are closely Related. In both cases a selection of model parameter values which is thought to be reasonable (or best) relative to some criterion applied to data is involved. In one sense, both procedures lead to identical outcomes.”

14. Instituto Nacional de Estadística y Censos (INDEC). https://www.indec.gob.ar/

15. We could not perform the numerical simulation exercise of reducing only taxes. This would cause the model not to converge, i.e., the deficit would increase indefinitely, violating the transversality condition of the model.

16. If $X_t$ and $Y_t$ are cointegrated, then the DOLS estimator is efficient in large samples, and statistical inferences about $\theta$ and the $\delta$‘s based on HAC standard errors are valid.

17. Others found that the results depend on the sample considered (Ilzetzki and Jin 2021) and on country characteristics, like the stage of economic development or the timing of the fiscal shock (see, for example, Kim 2015; Forni and Gambetti 2016; Miyamoto, Nguyen, and Sheremirov 2019; Boehm 2020;; Lambertini and Proebsting 2020).

18. Strictly speaking, the real exchange rate is another relative price, and it is determined in general equilibrium along with all other relative prices.