A New, Simple SFC Open Economy Framework

ABSTRACT

The paper presents a simple Stock-Flow Consistent open economy model with flexible exchange rates. It can reproduce the same dynamics and results of the flexible exchange rates ‘benchmark’ model by Godley and Lavoie [2007b. Monetary Economics: An Integrated Approach to Credit, Money, Income, Production and Wealth. London, UK: Palgrave MacMillan]. The latter is considered the ‘centre of gravity’ of SFC open economy literature. Yet the new model uses only one-third of the equations of the original one and features a different mechanism of determination of the exchange rate. Its small size and its flexibility make the model suitable both for didactic purposes and for extensions with further building blocks to address a variety of research topics.

KEYWORDS:

• Two-country model
• balance of payments
• exchange rates

JEL CODES:

• E12
• F310
• F320

Acknowledgements

I wish to thank Gary Dymski, Giuseppe Fontana, Marco Veronese Passarella, Luca Zamparelli and Gennaro Zezza for helpful comments and suggestions. I would also like to thank three anonymous referees and the Editor who provided useful and insightful criticism and suggestions. The usual disclaimers apply.

Disclosure Statement

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

Notes

1 Several SFC models have analysed the imbalances of the euro areas from a sectorial balance perspective. Among them, the ones presented in Godley and Lavoie 2007a; Ehnts 2013; Ioannou 2018.

2 The model in Godley and Lavoie (2007b) is made of 91 equations, but there are some gaps in numbering (for instance equations OPENFLEX 12.73 and 12.74 do not exist). The final number also depends on optional equations that may or may not be included.

3 In Godley and Lavoie (2007b) the simplest model presented is called ‘Model SIM (for simplest)’ (Godley and Lavoie 2007b, p. 58). That is a closed economy model.

4 Of course, the price of a currency must be interpreted as a price against another currency: in other words, the exchange rate against a foreign currency.

5 Different ‘varieties’ of disposable income are commonly used in SFC models for the consumption equations. In the original OPENFLEX these equations feature the expected Haig-Simons disposable income (see OPENFLEX 12.37 and 12.38 in Godley and Lavoie 2007b, p. 456), which is given by the average of the current Haig-Simons disposable income and the same variable with one lag. To limit the number of unknowns, in the OPENSIME we refrained from any expectation modelling and resorted to the standard Haig-Simons disposable income. Equations 11 and 12 are further simplified in experiment 3 (see Section Four).

6 For the sake of simplicity and to keep the analogy with the OPENFLEX model the income variable is not lagged (the lag would have created problems when the model is initially run with all starting values at 0 except the exchange rate: the natural log of 0 is undefined). However, the total factor income does not include the capital gains/loss linked to the depreciation/appreciation of the domestic currency. Therefore, it is quite isolated from the shock which is affecting the currency in the same period.

7 The Nobel Memorial Lecture of James Tobin (1982) can be considered the first ‘manifesto’ of the SFC approach to macroeconomic modelling. Among the principal features that differentiate Tobin’s framework from the standard macro-model of the time, there is the inclusion of ‘several assets and rates of return’ (Tobin 1982, p. 172). Here the number of assets is reduced to the minimum (6 in total: 3 types for each country). But the structure of the model is set in line with Tobin (1969) and in a way that it can incorporate as many additional assets and corresponding returns as desired by the modeller.

8 Tobin’s vertical constraints impose that the vertical sum of the first $\lambda$ for each country/households group is 1 and the vertical sums of the remaining $\lambda$s attached to the returns’ variables are 0.

9 The possibility for this complete symmetry to be given rests also on the fact that money has actually its own ‘rate of return’, which is assumed to be zero in the OPENSIME model.

10 http://gennaro.zezza.it/software/eviews/gl2006.php. This is a ‘translation’ in Eviews made by Gennaro Zezza of the original code written by Wynne Godley in Modler.

11 The supply of money (M1) is generally defined as the sum of cash money in circulation plus current bank deposits. The OPESIME model does not include commercial banks: it does not feature any variable for commercial bank deposits and for deposits of banks at the central bank (reserves). Consequently, the monetary base (M0) is made only of cash money. Since the money supply (M1) is only made by cash money too, the two measures of narrow money (M0 and M1) coincide. For convenience - and for consistency with the terminology used in Godley and Lavoie 2007b - we have used the expression ‘money supply’ to refer to cash money in this paper.

12 The variables in the equations are actually the supplies of assets, but we have seen with equation 23–26 they correspond to the demand of assets (the use of the supply variables just ensures currency consistency).

13 From this point of view the euro area could be modelled as a single ‘country’, given the existence of a single currency and a single central bank.

14 Of course, this is true for the economic dynamics that can be represented through the limited number of variables of the OPENSIME model. As no system of prices is embedded in the OPENSIME model, the ‘price-related dynamics’ of the OPENFLEX model cannot be replicated.

This also suggests that the results of other models based on the OPENFLEX’s closure (see Section One: ‘Introduction’) could be replicated with a new version of those models based on the OPENSIME’s closure. Indeed, in the code of Carnevali et al. (2020)—see section Five of this paper—both closures have been embedded, and the modeller can select which one should be used (the selection is made through a ‘dummy’ variable which activates and deactivates a handful of equations related to the two closures). The results of the experiments are not affected by the closure (although the OPENFLEX’s closure appears to be more resistant to ‘very big shocks’ as it is computationally simpler to solve by the software, given the high ‘simultaneity’ of variables in the exchange rate equation of the OPENSIME. Note that we are just referring to the closures: the code of Carnevali at al. 2020 has been built on the OPENSIME’s structure, not the OPENFLEX’s).

15 Obviously, as far as there are equivalent equations, variables and parameters in the two models. 

16 Sensitivity test has been conducted to prove this point. The stability of the results of the OPENSIME model comes with no surprise given its simplicity. And this another of its strength: the model cannot be suspected to carry results due to a tailor-made calibration. That is not always the case with larger SFC model, where it can happen that some parameters must be tweaked to ‘make the model run’. In the OPENSIME model the path to the new stable state following a shock does depend on the parameters of the import and export equations. However, different (reasonable) values for those parameters can alter the speed of the adjustment, not the fundamental dynamics of the model. 

17 The distinction between real and nominal variables applies only to the OPENFLEX model, given the fact that the OPENSIME model assumes fixed price and real and nominal variables coincide.

18 Exchange rates expectations are not included in the OPENSIME due to the effort to keep the model as simple as possible. If one would like to incorporate them, an additional variable should be added to equations 17-20. Then a theory to model the expectations should be introduced, like for instance in Lavoie and Daigle (2011).

19 Obviously, if one does not want to modify the OPENSIME model, the same kind of experiment can still be conducted, as far as the magnitude of the shock is scaled down.

20 In equation 14 (UK import) the exchange rate is lagged. If UK nominal import was given by real import multiplied by import price, it would be higher in value due to the depreciation of the pound. The fact that we ruled out prices from import and export equations—and from the model as a whole—produces a ‘neutralisation’ of the sterling depreciation on UK current account just after the shock. In other words, no typical ‘J-Curve alike’ phenomenon emerges. That’s another reason why the UK current account can turn positive almost immediately; consequently, the US external position is in deficit right from the beginning.