Foreign debt in a simple Stock-Flow Consistent Kaleckian growth model

Transcription

1 Foreign debt in a simple Stock-Flow Consistent Kaleckian growth model Pablo Gabriel Bortz Servaas Storm Paper presented at the Research Network Macroeconomics and Macroeconomic Policies (FMM) Conference Public debt, financial regulation, and income distribution Berlin, October Ph.D. Candidate, Delft University of Technology. Contact: p.g.bortz@tudelft.nl Assistant Professor, Delft University of Technology. Contact: s.t.h.storm@tudelft.nl

2 2 Abstract 1 We present an open economy growth model, using the Stock-Flow Consistent (SFC) methodology. We integrate international financial flows consideration into a demand-led growth model, with Kaleckian features. We investigate the effects and interactions that these inflows (and outflows) have on income distribution, foreign debt (both public and private), fiscal and monetary policy, both under fixed and freely floating exchange rate regimes. JEL Code: E12, E24, E31, E44, F32, F We are especially grateful to Marc Lavoie, who has been a permanent (and stoical) source of consult. At various stages of this research, we have benefited from the useful comments and suggestions of Fabián Amico, Amit Bhaduri, Eduardo Crespo, Adriana Diaz Arias, Amitava Dutt, Germán Feldman Alejandro Fiorito, the late Wynne Godley, Alfred Kleinknecht, Gustavo Murga, Ro Naastepad, Luis Rajuán and Manuela Robba. Obviously, none of the persons mentioned above are to be held responsible for the mistakes made in this paper, which are entirely our responsibility. We also thank the financial support of the NWO.

3 3 I. Introduction The stock of financial assets in the world has increased several times more than world GDP (Palma: 2009, p. 834). While most of financial flows are concentrated between industrialized countries, the share that goes and comes from developing countries has a substantial effect in these countries, due to their relatively small size, their immature financial development and to the fact that they have opened themselves (at least, many of them) to try to increase their attractiveness for foreign financial investors. As it is well known and documented, this financial opening has led in several occasions to crises in the balance of payments, of several types, in different circumstances 2. Though attracting much less attention unfortunately, around the same period (at least from early 80s, but with some pieces appearing in the 70s) there has been a development of models in the heterodox tradition in economics that tried to analyze economic growth from an effective demand point of view, labeled as Kaleckian 3. But in spite of the great amount of features that these models have incorporated (the list of works mentioned in footnote three is just the tip of an iceberg), in our knowledge there have not been articles that include international financial flows as a relevant factor in the economic performance, with the exception of Dutt (1990), Burkett and Dutt (199) and Bhaduri (2004) 4. Most of the models (for instance, Blecker (2002)) have been limited to treatments of the trade balance. This paper tries to make a first step in that direction, capturing some effects caused by the openness of the financial account. We do so by resorting to the the Stock-Flow-Consistent (SFC) Modelling tradition of Wynne Godley and Marc Lavoie (2007) in which the income accounting has no black holes (as Lance Taylor 2008, p. 639 writes). An income or financial inflow to an institutional sector (household, government, banks, and the rest of the world) is made up of outflow(s) from other sectors. The flows are cumulated over time so that the model is stock-flow consistent. The effects on wealth of gains or losses in asset prices are carefully accounted for, and so are differential inflation rates in prices for goods and services.. The stylized model (which is still in an explorative stage) features two economies (the USA and Argentina, making its first appearance in the SFC literature). The main innovation of this paper with regard to SFC models, as far as we know, is that we allow the Argentineans government and firms to borrow in international markets in a currency different from their own. We believe that this subject is becoming more and more relevant in an international context where some countries manage the value of their currencies and others let them float, leading to what the Director of the IMF has denominated as a currency war, something not very different from what happened in the early 30s, and we all know how that ended. 2 Just to mention one work that deals with many of these eposides, see Arestis (2006). For analysis and comparisons of different crises, especially during the 80s and 90s, see Kregel (1998) and Palma (2004), among many other articles. 3 The early story of these models is told in Lavoie (1995). Hein and Vogel (2008) do a summary of many theoretical and empirical articles in this tradition. Finally, see also Setterfield (2010) for some of the latest developments. 4 A recent study exploring somewhat similar issues is Cripps, Izurieta and Singh (2011).

4 4 The focus will be on what happens to the developing country, leaving aside the effects on USA, as long as they are not necessary for the presentation of the argument. The rest of the paper is as follows. Section II outlines the model. Section III presents and interprets the results from our simulation experiments. Section IV draws out the main (policy) conclusions within the context of the current debates on rebalancing. II. The model A. Overall description The model comprises two economies ( USA and Argentina, making its first appearance in the SFC literature as a country s name) with five sectors in both of them: households, firms (which are vertically integrated, so that their sole inputs are wages and imports), banks, government and the central bank. Even though the latter is an institution within the public sector, we find more convenient to distinguish its operations from the ones of the central administration. Both countries exchange goods (each country produces one good, useful both for consumption and investment purposes) and financial assets. The next two matrices present the accounting of stocks and flows that occur in our model. First, we present the balance sheet of the model, with assets carrying a plus sign, liabilities and net wealth a minus sign. [INSERT TABLE 1 HERE] Households have accumulated wealth in the form of a range of financial assets: cash (H), deposits (M), and Treasury bills. In the model, there are three kinds of bills, which are expressed in the following way: bills issued by the USA government, denominated in dollars ( ); and two types of bills issued by the Argentinean government: one kind denominated in pesos ( $, which can be acquired both by USA and Argentinean households, Argentinean banks and the Argentinean central bank), and another denominated in dollars ( $ ), which can only be purchased by USA investors. Within each country, all items are expressed in their currency, since the relevant assets $ (and liabilities) are transformed by the exchange rate. For instance, represents the holdings of Argentinean bills denominated in dollars (as shown by the supra index $u ) by American 5 households (shown by the sub index hu ). Since the bills are denominated in dollars, no transformation is needed here. But the holdings of bills denominated in pesos by these same agents ( $.) requires to be multiplied by the exchange rate, in order to be expressed in dollars. The same happens with assets (and liabilities) held by Argentineans, issued in dollars. Notice that while the US Central Bank does not have any foreign reserves, the Argentinean Central Bank does, in the form of US Treasury Bills (.), translated into pesos by the exchange rate of that country. Firms, in turn, have capital stock as an asset and loans as liability. The difference between the two represents the net wealth of firms. An innovation has been introduced here: 5 By Americans, we understand in this work the residents of the USA, what in Spanish are called estadounidenses, literally, from the United States. Some Mexican friends have told us that also this word is incorrect, but we want to avoid further nationalistic susceptibilities.

5 5 we allow Argentinean firms to borrow in two separate banking systems: in their home country (represented by ) and in the USA (represented by, loans granted by American Banks and therefore denominated in dollars). Banks have deposits as their sole liability, and loans and Treasury bills as their assets. In the eventual case that loans surpass the amount of deposits, bills holdings will be negative, representing a kind of advances from the government. Their net wealth is zero, since it is assumed that they do not make profits (as will be explained later, the interest rate on bills, loans and deposits will be equal). Finally, the counterpart of the central bank s liability (cash) is its holding of Treasury bills. Reserves are a kind of net wealth of the Argentinean central bank (more on this later). [INSERT TABLE 2 HERE] This is the transaction matrix that tracts all the flows that occur in one period. We have decided to split the columns of firms and banks into two, in order to highlight the difference between current transactions (which do not affect the capital of the firm) and capital transactions, (which do). An important caveat should be made here: a plus represents a source of funds, and a minus represents a use. For instance, when households buy cash (investing in it some of their wealth) this is represented with a minus sign in the households column (because it s a use of their wealth) and with a plus in the column of the central bank, since it represents a source of funds for it. The first five rows represent the components of GDP: consumption, government expenditure, investment, exports and imports. Investment is recorded with a plus sign in the current account of firms, since it s a source of funds; and with a negative sign in the capital account, assuming that firms buy their capital stock to themselves. Only firms are allowed to import, using them as an input in their production process. Notice that, since within each country all expressed in their own currency, the only case in which (imports of the USA) equals is when the exchange rate is one. Firms pay. as the wage bill to households, who pay income taxes (the only kind of taxes in this model) to the government. Households also receive interest payment arising from their deposits at banks and their holdings of bills (so do banks and central banks). These payments are determined by the interest rate and the amount of deposits (or bills), both of the previous period, though in the relevant cases, these payments are translated into the corresponding currency using the exchange rate of the current period (as expressed, for $ instance, in (). ()., the payment that American households receive on their past holdings of Argentinean bills denominated in pesos). The same happens with interest payments on loans. Central banks, on their turn, give back to their government the interest payment they received on the bills issued by them, their profits. That is straightforward in the case of the Fed, but it s not so in the case of the Argentinean central bank: it returns to the Argentinean government only the interest payments on Argentinean bills, not on the USA bills. As will be explained later, this will add to their reserves (as seen with. ), which will increase through time, even in the flexible exchange rate regime. What is left for households after they have received their wages and interest payment and paid their taxes (and capital gains, as will be shown later), is their disposable income. From that, we deduct consumption and get the increase in their wealth. This is invested in cash, deposits, and bills, being a source of funds for the central bank, commercial banks and

6 6 governments, respectively. Given the accounting constraint that everything comes from somewhere and goes to somewhere else, the increase in circulation (a liability in the central bank s balance sheet) is matched with an increase in its holding of government bills. Firms finance their investment with their profits and loans borrowed to banks. Like we said, the excess of deposits over loans add to banks holdings of government debt. B. List of equations 1. Identities and National Accounts The next box presents the equations related to national accounts and identities, distinguishing between real and nominal values: real values are written in lower-case letters, nominal values in upper-case. Real sales are the sum of real consumption, investment, government expenditure and exports. This number is multiplied by the price of sales to achieve nominal sales. Real GDP is equal to real sales minus real imports, and something similar happens with nominal GDP. Box 1 1) = Real sales in USA 2) = Real sales in Argentina 3) =. Nominal sales in USA 4) =. Nominal sales in Argentina 5) = Real GDP in USA 6) = Real GDP in Argentina 7) = Nominal GDP in USA 8) = Nominal GDP in Argentina 9) = 10) = GDP price deflator in USA GDP price deflator in Argentina 11) = Real exports in USA 12) = Real exports in Argentina 13) =. Nominal imports in USA 14) =. Nominal imports in Argentina 15) =. Nominal exports in USA 16) =. Nominal exports in Argentina 17) =. Nominal consumption in USA 18) =. Nominal consumption in Argentina 19) =. Nominal investment in USA 20) =. Nominal investment in Argentina 21) =. Nominal public expenditure in USA 22) =. Nominal public expenditure in Argentina

7 7 Equations (11)-(12) says that real exports of one country are equal to the real imports of the other country. Combined with equations (15) and (16), they convey the following (Keynesian) basic assumption: the importer country determines quantities, the exporter determines prices. Other alternatives could be: that one country determines the prices of both its imports and its exports (for the exporter of the other country, it would just mean that its export prices are exogenous), which is not an unrealistic assumption for some sort of commodities (especially raw materials and agricultural products), though not for manufacture products, in general; or a rule that equalizes both prices (say, they adjust to some profit rate related to the interest rate of one country like in Sraffian models, etc). Equations (17) to (22) represent just the nominal values of the components of GDP, while equations (9) and (10) show the GDP price deflator. 2. Households Box 2 23) =. + (). () + (). () $ + (). () 24) = $. + (). () + (). () + (). (). $.+ (). () Regular income in USA Regular income in Argentina 25) =. Tax payments in USA 26) =. Tax payments in Argentina 27) = $ +. () 28) = +. () 29) = (). 30) = (). 31) = () +. () () 32) = () +. () () Nominal disposable income in USA Nominal disposable income in Argentina Real disposable income in USA Real disposable income in Argentina Expected real disposable income in USA Expected real disposable income in Argentina 33) = () + Nominal household wealth in USA 34) = () + Nominal household wealth in Argentina 35) = Real household wealth in USA 36) = Real household wealth in Argentina 37) =. +. () Real consumption in USA 38) =. +. () Real consumption in Argentina

8 8 Now we turn to the equations relating to the behavior of households, based on Godley & Lavoie (2007, chapter 12). Regular income for households is composed of the wage bill and the interest payments on the financial assets they hold, calculated on the interest rate and assets hold at the end of the previous period. On that regular income, they pay taxes, so that nominal disposable income is equal to the regular income minus taxes, plus capital gains (or minus capital loses) arising from changes in the values of foreign bills. These changes can only be brought about by the movement of the exchange rate. Real disposable income, instead, has to take account of real losses on the financial wealth arising from inflation (or gains arising from deflation), and that is what is shown in equation (29)-(30). Equations (33)-(36) express nominal and real wealth of households, respectively. Equations (37)-(38) say that households decide their consumption in real terms. Consumption depends on expected disposable income and on the accumulated wealth at the end of last period. The way we model expectations equations (31)-(32) is simple: current expected disposable income is equal to past disposable income plus (or minus) an adjustment term correcting for previous mistakes. Therefore, like we said in the previous paragraph, real losses (gains) on the financial wealth of households due to a rise (fall) in prices have an influence in the consumption of this sector; they are not blind to price changes. After the decision about how much to consume has been made, households decide where they are going to allocate their savings, that is, their wealth. As stated previously, American households can decide among a set of five financial assets (cash, deposits, American bills, and two types of Argentinean bills), while Argentinean households have only four options, because the US only issues one type of bills. Households make their allocation decision in terms of their own currency, not in the other s. For instance, Argentinean households demand American bills equivalent to a certain amount in pesos,. SFC models usually adopt Tobinesque rules for modeling assets allocation, but in order to keep the model (a little bit) simple(r), we have chosen to assign a constant proportion to each asset, making some experiments later on different preferences by investors. This is presented in box 3. Equations (43A) and (47A) do not enter into the model because they would cause overdetermination of the system. A suffix d, at the end of these variables, means that they represent a demand, and a suffix s represents supply. Box 3 39) = Demand for deposits in USA banks 40) = Demand for bills issued by the USA 41) $ = Demand for bills issued by Argentina denominated in pesos 42) $ = Demand for bills issued by Argentina denominated in dollars 43A) = Demand for cash in USA 43) = $ $ 44) = Demand for deposits in Argentinean banks 45) $ = Demand for bills issued by Argentina denominated in pesos

9 9 46) = Demand for bills issued by USA 47A) = Demand for cash in Argentina 47) = $ 3. Firms Since this is a demand-led growth model, with no supply constraints (either labor or capacity), the level of production of firms is determined by demand, and we have already explained one of its components, namely consumption. It s time to explain the determination of prices, foreign trade and investment. All these items are presented in Box 4. Box 4 48) = (1+ ).(. + ) Sales prices in USA 49) = (1+ ).(. + ) Sales prices in Argentina 50) =. () Real imports in USA. () 51) =. () Real imports in Argentina. () 52) =. Import prices in USA 53) =. Import prices in Argentina 54) =. Nominal capital stock in USA 55) =. Nominal capital stock in Argentina 56) = ().(1+ ) Real capital stock in USA 57) = ().(1+ ) Real capital stock in Argentina 58) = (). Real investment in USA 59) = (). Real investment in Argentina 60) = () 61) = () Capacity utilization rate in USA Capacity utilization rate in Argentina 62) = Employment level in USA 63) = Employment level in Argentina 64) = +. (). Growth rate of capital stock in USA 65) = +. ()... Growth rate of capital stock in Argentina 66) =. (). () 67) =. (). () (). (). 68) = 69) =. Profits of American firms Profits of Argentinean firms Net wealth of American firms Net wealth of Argentinean firms

10 10 Prices are set by firms according to a simple rule: it s a mark up over unitary costs. Since firms are vertically integrated in each country, costs are made of the wage bill and imports (remember that firms are the only sector that has foreign trade in this model). Equations (50)-(51) show a standard import demand function, where the explanatory variables are relative prices and the real GDP, with ε 1 and µ 1 being the prices elasticities and ε 2 and µ 2 the income elasticities of imports. Prices of imports are depicted in equations (52) and (53). Equations (54) to (63) present the nominal and real capital stock, that grows at a rate equal to and ; real investment; the capacity utilization rate and the level of employment, which is equal to real GDP divided by labor productivity (an exogenous parameter, in this paper). For the growth rate of capital we have chosen a very simple investment function, not because we necessarily agree with it 6, but because it has been proved acceptable in achieving the baseline, and because it lends itself to useful experiments related to liquidity preference and monetary policy (though we did not conduct strict simulations of changes in monetary policy, for reasons that will be explained later). We include the capacity utilization rate (as a proxy of the accelerator principle) and the interest rate, as a proxy for the cost of finance. We choose to include the real interest rate in the investment function (with a low coefficient), though in the case of Argentina we made a distinction between domestic sources of loans (for which the real interest rate is the relevant indicator) and foreign borrowing (in which case the nominal interest rate on American loans times the exchange rate is the convenient variable). Profits are a residual over costs and interest payments (equations (66)-(67)). Finally, net wealth is the difference between capital stock and loans. Implicitly, it is assumed that firms invest in capital stocks all their profits, since there is no other asset to match the net wealth. Since we are talking about borrowing and finance, let s present the equations related to these subjects. We adopt an endogenous money approach, so that banks grant loans on demand. There are four equations that explain the behavior of firms on these matters: Box 5 70) = () + Demand of loans by American firms 71) = () 72) = + (). + Total demand of loans by Argentinean firms Demand of loans by Argentinean firms to Argentinean banks 73) = Demand of loans by Argentinean firms to American banks Equations (70) and (71) say that firms demand loans (which last only one period) by an amount equal to their previous debt, plus their investment plans minus their profits. American firms only borrow in the domestic banking system; Argentinean firms do so in both loan markets, borrowing a proportion δ to Argentinean banks (eq. 72) and a proportion (1 δ) in American banks. The outstanding amount of debt in pesos is affected, in the case of Argentinean firms, by movements in the exchange rate, since loans granted by US banks are denominated in dollars. 6 In particular, it has already been shown that investment is not a monotonous function of the interest rate, both theoretically (with Keynes and also with the Cambridge controversies, among other exponents) and empirically (for a neoclassical example, see M. Taylor (1999))

11 11 4. Banks Like we said, working in an endogenous money framework, banks grant loans on demand to creditworthy borrowers, a qualification that in our model comprises all firms of both countries. They also accept deposits, and the difference between the latter and the loans they give is invested in domestic bills. All this is shown in Box 6. Box 6 74) = 75) = 76) =. 77) = 78) $ = Supply of loans to American firms Supply of loans by Argentinean banks to Argentinean firms Supply of loans by American banks to Argentinean firms Demand of bills by American banks Demand of bills by Argentinean banks 79) = Supply of deposits by American banks 80) = Supply of deposits by Argentinean banks 5. Government Governments see their indebtedness increases as its total expenditure exceeds its income. In general terms, governments give bills on demand (eq. (84) and (86)-(93)), with one exception: the USA government has a rationing mechanism that depends on the choice of the exchange rate regime closure. Equations (85FL) and (88FL) correspond to a flexible exchange rate closure; we ll explain later why it is so. We have chosen real government expenditure as the autonomous variable, which grows at a fixed rate of 3%. Nominal government expenditure is an endogenous variable affected by changes in the price level. To save space, the only additional thing worth to mention is the fact that the interest rate on bills each government pays to its own central bank, returns to the same government as a sort of profits of the central bank (last term of equation (81) and (82)). In the case of the interest payments that the American government gives to the Argentina s central bank, these add to the existing amount of foreign reserves, increasing the demand of American bills 7. However, this supply does not respond to the behavior of the balance of payments, as the fixed exchange rate regime does. All these equations are shown in box 7. 7 This feature in an otherwise typical flexible exchange rate closure, is named Mathieu Lequain s closure, as explained in the Appendix to chapter 12 of G&L. We share the view that this feature does not change the essential characteristics of a flexible exchange rate model. We want to capture with it a trend that we identify in recent times: even though formally central banks do not interfere in exchange rate markets (with the exception of China), in practice they have intervened in order to defend a parity they want, even if it s not announced. But there are cases where foreign reserves in dollars have increased (two examples coming to mind are Brazil and the UK) without one having the chance to identify any definitive orientation in the exchange rate policy, not to say contradictory trends such as a systematic appreciation of the currency like Brazil.

12 12 Box 7 81) = () + + (). () (). () 82) $ = $ + $ $ ) = $ ().(1+ ) 84) = $ + ().1+ (). + $ (). () 85FL) = 86) = Supply of bills of the USA government Supply of bills in pesos Total supply of bills of the Argentinean government Supply of USA bills to USA residents Supply of USA bills to Argentinean residents Supply of USA bills to USA banks 87) = Supply of USA bills to the USA central bank 88FL) = ().(1 + () ) Supply of USA bills to the Argentinean central bank 89) $ = Supply of Argentinean bills in pesos to Argentinean residents 90) =. Supply of Argentinean bills in pesos to USA residents 91) $ = Supply of Argentinean bills to Argentinean banks 92) $ $ = Supply of Argentinean bills to the Argentinean central bank 93) =. Supply of Argentinean bills in dollars to USA residents 94) = ().(1+ ) Real government expenditure in USA 95) = ().(1+ ) Real government expenditure in Argentina 6. Central Banks, interest rates, inflation and flexible exchange rate Finally, we present the equations relating to the behavior of the central bank and balance of payments. These are shown in box 8. Box 8 96) = Supply of cash in USA 97) = Supply of cash in Argentina 98) = Demand of bills by the USA central bank 99) $ 100) = Demand of bills in pesos by the Argentinean central bank =. Demand of bills in dollars by the Argentinean central bank Profits of the American central bank 101) = (). () 102) = $ (). () Profits of the Argentinean central bank

13 13 103) = Nominal interest rate on loans in dollars 104) = Interest rates on deposits in dollars 105) = Interest rates on bills in dollars 106) = Nominal interest rate on loans in pesos 107) = Interest rates on deposits in pesos 108) = 109) = (1+ ) (1+ ) 110) = (1+ ) (1+ ) 111) = ( ) Interest rate on bills in pesos Real interest rate on loans in dollars Real interest rate on loans in pesos Inflation rate in USA () 112) = ( ) () 113FL) = 114) = 1 Inflation rate in Argentina Exchange rate in Argentina Exchange rate in USA Equations (96)-(99) say that the monetary base is endogenous; while the demand of bills by the Argentinean central bank follows its supply (eq. (100)). Equations (101)-(102) show the profits of central banks, while equations (103)-(112) express interest rates in both economies. In a SFC model, each endogenous variable appears in the left-hand side in only one equation. We have shown what determines the demand for American bills by Argentinean households (eq. 46) and what determines, in this closure, the supply of these bills (eq. (85FL)), which are a residual. We let to the exchange rate the role of equilibrating both quantities, in equation (113FL). However, this does not mean that the exchange rate is determined solely in this market ; on the contrary, its value satisfies each equation in which it appears, and depends on them also. In general terms, it will be seen that it depends on the financial position of all the sectors in the model. 7. A fixed exchange closure, and the balance of payments Box 9 85FX) = In order to present a fixed exchange rate closure, we only need to change 3 equations:. 88FX) = Supply of USA bills to Argentinean residents Supply of USA bills to the Argentinean central bank 113FX) = Exchange rate in Argentina In equation (85FX), we let the supply of bills to Argentinean households follow its demand (so that no adjustment in prices is required there), while taking the supply of bills to

14 14 the Argentinean central bank as a residual (eq. (88FX)), the demand adjusting to this number in equation (100). The last equation explicitly says that the exchange rate in Argentina is constant. There are other alternative closures for a fixed exchange rate regime, revised in chapter 12 off G&L, but we decided to adopt the main, conventional form. We finish the presentation of the equations of the model with the balance of payment. Remember that there are expressed in the currency of their own country, so the current (and capital) account of Argentina and the US do not necessarily sum up to zero, though they do have opposite sign. Box ) = $ $ + (). ().+ (). () + (). () (). () (). () ) 116) = + (). () + (). $ $ (). () (). (). (). (). 117) = + $. $ 118) = $ + $.... Current account of USA Current account of Argentina Capital Account of USA Capital account of Argentina III. Simulations A. The baseline A word should be said about the baseline before moving to the results of the simulations we performed. The initial values of the variables and the values of the parameters are listed in the appendix. The two economies have the same initial values for endogenous and exogenous variables and parameters, in both closures. We stress the fact that, in the portfolio composition, even though the weight of cash and deposits is high, we still retain a predominance of bills, and a home bias, expressed in two ways: a much greater predominance of domestic bills in the portfolio, and in the case of US households, a preference of foreign bills denominated in dollars instead of pesos. In the case of the fixed exchange regime, both economies growth at the same rate (3%), with the same values for the wealth of households and firms, and with a balanced current account. There is a slight difference in the behavior of banks: American banks face a greater demand for loans, so up to a point they have to ask advances to the central government because they run out of bills. This problem is bigger in the flexible ER regime than in the fixed. In the case of the flexible ER closure, even though both economies have the same initial values of parameters and variables, they grow at a slightly different rate, though in a converging way. In the last period of the baseline, the accumulation rate in Argentina reaches 2.99%, while the USA it is 3.03%. The level of real output is 5.5% greater in the latter than in the former. The peso experiences in the long term an appreciation, up to 13% (remember that in the fixed ER closure we observe an increase in foreign reserves, even though the current

15 15 account is in balance), and the current account is negative, though slightly converging to balance. In our view, movements in the exchange rate cause disequilibrium in the trade balance, thus a greater rate of growth in foreign demand in USA. The cause of this movement in the exchange rate is the fact that there is a shortage of private savings in the US, or in a better expression a shortage of funding, compared to the requirements of the government and the evolution of nominal wealth (and demand of American bills) in Argentina. However, like expressed previously, the differences are reducing their size: in the last year of the period, the trade deficit represented in Argentina less than 0.15% of its nominal GDP, while the current account deficit was less than 0.5%. What this results showis that a careful analysis of the sectoral financial balance is a requirement in order to assess the evolution of the exchange rate, and several contradictory situations will appear to support this position. B. The paradox of thrift If we claim that our model is a demand-led growth model, at least we should be able to observe the paradox of thrift in our simulations, and that s exactly what we see. Figure 1 presents the evolution of real GDP under both closures, when we raise the propensity to consume out of (expected) disposable income of Argentinean households from 0.8 to 0.85 (a 6.25% higher). We have two counteracting effects acting on the consumption demand because, even though households are more prone to consume out of income, the change in the parameters in the consumption function brings about a decline in the long term relation between the wealth of households and disposable income, i.e. it diminishes (G&L: 2007, p. 75), simply because families decide to consume more and save less. We see, therefore, that both nominal and real wealth of Argentinean households diminish relative to the baseline. In the overall, however, the first positive impact is more relevant. [INSERT FIGURE 1 HERE] Compared to the baseline, there is an increase in real GDP (and employment, since with a given productivity both variables go hand in hand), which is stronger in the short term and later on diminishes as the trade balance deteriorates. As a matter of fact, we also find the expected deterioration in the current account, as well as an important loss of reserves in Argentina (in the fixed ER regime) or a depreciation of the peso up to 21% in some moments, though only 10% and decreasing in the end (in the flexible ER). However, the exchange rate (or reserves) does not necessarily move in the same direction as the current account, as we will see later. Notice also that the positive effect on Argentina s employment (and negative on USA) is greater in the flexible ER closure (3.5% higher vs 1.9% in the fixed regime): even though the depreciation of the peso still produces some inflation, the latter is not sufficient to offset the positive effect of the former (the price level at the end of the simulation is only 1.3% higher than the baseline), giving a stimulus the trade balance, or at least lowering the damages. Figure 2 presents the current account of Argentina, expressed in dollars. [INSERT FIGURE 2 HERE] As regards Argentinean public debt, our results show a substantial decline in its level compared to the baseline, supporting therefore the claim of G&L (2007, p. 96-7): a higher propensity to consume is linked, given the level of activity, with a lower budget deficit and

16 16 lower public debt. However, foreign public debt represents a greater share of government liabilities in that country, because we do not observe the same fall in the wealth of American households. Figure 3 presents the debt/nominal GDP ratio of Argentina compared to the baseline. Keep in mind that, as deposits also go down, Argentinean banks demand a lower amount of bills, adding another explanation to the falling debt/income ratio. C. Distribution [INSERT FIGURE 3 HERE] We move now to the analysis of the impact of changes in distributional parameters, starting by a 10% increase in the productivity level of Argentinean firms. As in the previous simulation, one has to distinguish (very) short run and long run effects. The first and most obvious consequence of this change is that in the first period the employment level in Argentina falls 10%, as well as nominal disposable income (though a little bit less). However, this is not all: there is also a fall in prices, due to a lower wage bill 8, so that households have a real gain on their accumulated wealth, that more than compensates the fall in wage income: real disposable income rises 22% in the first period, under both exchange rate regimes 9. In other words: households are hurt because they are workers, but they also benefit because they are rentists. Remember that in this model banks do not make profits. Since we assume that households decide their consumption in real terms, i.e. they are not blind to inflationary losses (or deflationary gains), in the next period real consumption of Argentinean households will increase, also real GDP, imports and profits. But this is a very short term impact. In the long term, nominal wealth will be lower in Argentina (and a little bit in USA, but for other reasons), but real wealth will be around the same level than the baseline (G&L: 2007, p. 295). The counterpart to a lower nominal wealth is a lower government debt, in Argentina. Deflation lowers government nominal expenditure, but taxes do not diminish that much: even more, since households consume more in the period following the change, income increases and so do taxes. In the end, the financial deficit of the government is lower, and therefore its indebtedness falls compared to the baseline. But the magnitude of the effects is different, according to the exchange rate regime. Figure 4a and 4b shows the evolution of real disposable income, the GDP and the employment level in Argentina, in the flexible and the fixed closure, respectively. [INSERT FIGURE 4a AND 4b HERE] You may think that deflation has all but negative effects in this model, but that is not so. The point is that, since firms cannot retain the gains in productivity, actually they are the ones hurt by the lower price level. In Figure 5, we show the profit share (after interest 8 But the fall is not a 10%, since only one output reduces its weight. In a case of a depreciation, imports also increase, but since they are only part of the cake, relative prices turn in favor of the country with the weaker currency. 9 In the steady state, with no mistaken expectations, there is a stable relation between wealth and disposable income, equal to: =., with = ( ) [ ], where is the growth rate of disposable income and consumption. With the parameters we have chosen and a growth rate of 3%, α 3 equals A real loss (or gain) in wealth has an enormous impact on disposable income.

17 17 payments) on real GDP and the ratio loans/nominal sales in Argentina, both compared to the baseline. The mass of profits (in nominal terms) is deflated by the GDP price deflator. As one can see, the share of profits declines as the ratio loans/sales increases. Another interesting factor is that firms are better off in a flexible exchange rate regime than in a fixed one. [INSERT FIGURE 5 HERE] The opposite happens with employment and output: the effect would be more positive to Argentina under a fixed exchange rate regime than under a flexible one. The reason is that a flexible exchange rate tends to attenuate the imbalances in the current account, which in this case favors Argentina. There are a number of reasons that lead together to an appreciation of the peso: a lower government deficit, a trade surplus, a greater nominal wealth in USA than in Argentina (relative to the baseline), all these features cause an appreciation of the peso so that in the end, real exports in both countries return to the values of the baseline, as well as real income. Instead, under a fixed exchange rate regime, what we see is a continuous increase in foreign reserves, which reinforces the surplus in the current account, without damaging Argentina s exports. Therefore, real GDP in Argentina (USA) rises (falls) relative to the baseline, even though there is greater unemployment. What we want to stress is that the effect on output (and the relative improvement in unemployment) is due to the effect on the balance of trade: as figure 4a shows, the effect on output in the flexible exchange rate case is negligible. Any coincidence with the situation in Europe is not casual. In our view there is no mechanism in the Eurozone that corrects (in an expansionary way) the difference in relative costs between the countries 10, so that the current account remains in a deficit. In the flexible exchange rate regime, Argentina even has a higher foreign debt as a share of its total debt, though its burden is reduced by the higher value of the peso. What happens in the case of a rise in the nominal wage of Argentinean workers? We raise Argentinean wages by 10% in 1960 (leaving them there for the rest of the period), and in many senses we observe compatible effects with what we have just described, but in the opposite direction. We have an initial rise in prices in both countries (since the product of one country is an input of the other), so that although regular and disposable nominal income go up (together with nominal wealth), real income falls due to the capital loss, affecting consumption in the next period (with a plus, since workers are correcting mistaken expectations of the previous period). In this case, households are favored because they are workers, but they are hurt because they are rentists. This fall in consumption has a multiplicity of effects. It lowers GDP, and therefore imports: in the initial periods after the shock, Argentina has a surplus in the trade and current account. We feel that this requires a somewhat more detailed exposition: The change occurs in In that year, there are no changes in the volume of goods traded, nor in any component of real output, since consumption, investment and exports are determined by past variables and government expenditure stays the same. But the nominal value of exports changes. In 1961 the effect of 10 We refer to costs and not productivity alone, because there has been a kind of convergence in that indicator, but changes in costs, particularly wages, have not allowed for instance to Greece to reduce the difference in competitiveness with Germany.

18 18 the fall in the GDP of Argentina is felt with more strength, explaining the surplus. The USA suffers collateral damage: its imports increase their price, producing a jump in the price level of 2.4% accumulated in the first 2 years (without having any increase in wages, of course). At the same time, its exports sunk due to the lower demand in Argentina, hurting its tax revenues and causing a slight increase in the budget deficit in a moment when its domestic funding is being reduced. So, in spite of the fact that nominal household wealth in Argentina rises relative to the same variable in the USA, we observe at these initial stages an appreciation of the peso. But later on we see the expected results: imports recover faster in Argentina than in USA (because of the change in relative prices), so that real income and consumption grow in USA more than in Argentina. But the exchange rate operates here in the opposite sense than in the previous simulation: a flexible exchange rate attenuates the imbalances in the trade balance (which in this case favors USA), up to the point that in the long term Argentinean imports (in dollars) are lower than the baseline. This helps to have almost the same level of employment in the long run in Argentina compared to the baseline. This is what figure 6 shows. The variable trade balance shows the deterioration in real trade or volume of goods traded, as percentage of real GDP. They are measured in the RHS axis. [INSERT FIGURE 6 HERE] On the contrary, under a fixed exchange rate regime Argentina has a permanent deficit in the trade balance (and the current account), which leads to a budget deficit (since the level of income and its tax revenues are lower than nominal government expenditure) and a lower level of employment. The loss in reserves is also an obvious effect of these movements. A not so obvious feature is that profits in Argentina rise compared to the baseline, in both regimes. Why? Well, wages can be transferred to prices without any counterattack on behalf of workers (we did not model the inflationary conflict), and added to the fact that there is also a slight increase in the price of imports, the GDP price deflator is somewhat higher than sales prices (which affect the price of capital goods), lowering the requirements for external funds and rising profits. We observe the opposite behavior of figure 5. A modeling of the distributive struggle would certainly change these conclusions. Some unexpected things happen, however, when we modify another distributive parameter in Argentina, the mark up rate 11. Given its low magnitude, its effects are more felt in the long run rather than in the short. For instance, a 20% increase in its value represents a jump in the price level of only 0.7%. Many results are compatible with what a Kaleckian model of growth would suggest: in the long term there is a lower level of GDP, a lower rate of growth, lower level of employment, all caused by a lower households disposable income and lower consumption. Figure 7a and 7b shows those variables. [INSERT FIGURES 7a AND 7b HERE] As a result, the current account and the trade balance of Argentina have a surplus, slightly greater in a flexible ER closure, because there is a depreciation of the peso, in spite of the fact that relative prices turn against Argentina (though we do acknowledge that with other 11 The mark up is really low, 3.5%. This is due to the relations we have chosen to have between wealth and disposable income, and to the fact that we wanted to have prices not very far from 1.

19 19 parameters the result might be different) In the fixed ER closure, Argentina faces a loss of reserves, in spite of the current account surplus. Comparing to the Kaleckian model of Hein & Vogel (2008, p. 486) and Bhaduri-Marglin (1990), in our model we need to distinguish carefully the nature of the exchange rate regime, but we can certainly observe that the effect of a rise in the mark up has more stagnationist features than the exhilarationist effects of a higher nominal wage. This is the opportunity in which we go into greater detail to explain why the exchange rate and the current account do not go necessarily hand in hand as expected. The explanation lies in the financial position of all the sectors in the economy. Strange as it sounds, there are some common features with what happened in the immediate aftermath of the crisis in 2008: instead of a fall in the dollar, what we witnessed was an increase in its value, when the whole world was collapsing. A somewhat similar mechanism can be used here. To start with, this is a case where there are not twin deficits: there is a foreign surplus together with a budget deficit, in Argentina. In the USA, however, there are actually twin deficits. Why is there a budget deficit in Argentina? Well, like we said there is a fall in household income that lowers tax revenues. This same fall in income triggers a positive trade balance, also helped by the fact that profits explode. The profit share on real GDP increases between 173% and 180% (depending on the exchange rate) from point to point. There is a great increase in domestic savings, even though real income has fallen. Firms do not invest more (actually, the real capital stock is a little bit lower than the baseline), instead they reduce their indebtedness both with Argentinean and American banks. As a result, even though the Argentinean public debt has increased due to the budget deficit, foreign debt has reduced its share. Something similar happens in the USA: the government finds a much eager demand for bills in the domestic market than in the foreign market, determining a fall in its supply to the rest of the world. Figure 8 shows the supply of loans to Argentinean firms by USA banks as a share of American deposits, the supply of American bills, and the evolution of the exchange rate. [INSERT FIGURE 8 HERE] What we extract of all this is that the exchange rate is not moved entirely by the current account (actually, we believe the opposite holds: it is the exchange rate that drives the current account), it is driven by the financial positions of all the sectors involved in the balance of payments: households, firms and the government. We have rejected several times in this paper the IMF view of the twin deficits process, but there is no one-size-fits-all corollary as to whether a budget deficit is a cause or consequence of a depreciated exchange rate, and the role of the current account in it. A detailed analysis of the situation is required in order to draw policy conclusions. The next simulation will be another opportunity to prove it. D. Fiscal policy It is time to look at the effects of higher public real expenditure. The experiments consisted in raising the growth rate of public real expenditure from 3% to 3.6%, in each country at the time. When we performed the simulations, we had convergence in the USA experiment, but in the Argentinean case we had values for 90 out of the 100 years of the

20 20 period sample. However, this was enough to observe that the results where analog to the other scenario, so we will focus our attention in the simulation where we increase the expenditure of the American government. To get the result of a higher expenditure of the Argentinean government, the reader just need to change the names of the countries; we will notice the only significant difference worth to mention. In the case of an increase in real expenditure in USA, we do observe in that country the twin deficits situation: a current account deterioration and a budget deficit, together with a greater real DGP in the US (compared to the baseline). However, the magnitude of the increase in the GDP, the deterioration of the current account and the impact on Argentina s performance depends on the choice of the ER. Under a fixed exchange rate regime, both countries enjoy the benefits of this larger US budget deficit, through the channel of foreign trade. The wealth of households rises in both countries, increasing the demand for public debt, though not completely at the same pace as the US deficit: the difference is absorbed by the Argentinean central bank, with a higher level of reserves. Would the ER regime be a flexible one, the story would be different. [INSERT FIGURE 9 HERE] Figure 9 presents the evolution of the Argentinean real GDP in the first case (fixed ER) and in the second (flexible ER). What is the cause of such divergent results? The answer lies in the ER: the opposite of the increase in reserves (in the fixed closure) is an appreciation of the peso 12, so big that it turns the trade balance into a deficit (for Argentina), lowering GDP and investment. If real consumption does not diminish that much, it s not due to some encouraging effect that one could expect in the real world: the reason is that the fall in nominal disposable income (which is even slightly greater than the fall in GDP, because it includes capital losses) is compensated by the fall in the price level. Not much to expect from there in the reality. So our results seem to support the policy measures advocated by Godley et al (2005), inasmuch as a flexible ER (or a lower dollar for that matter) is a stimulating factor for the US when its government adopts a fiscal expansion. They also provide basis for the ancient wisdom of the Chinese government who knows, perhaps ever since the Han dynasty, that it should not allow its exchange rate to float or to appreciate, increasing its foreign reserves instead in the necessary amount, however large it might seem. E. Dutch disease? This paper does not include simulations about strict changes in the monetary policy of each country. The reason to exclude them is that the usual transmition mechanisms have been omitted since the beginning: by frozen the composition of households portfolio, the demand of assets does not respond to movements in the relevant rate of return. We did not include exchange rate expectations; the sensitivity of investment to the interest rate is very low (with which we agree, specially with low interest rates, but not that much with important rises); and household consumption is positively affected by the interest rate, as much as it increases the 12 In the case of a rise of the Argentinean public real expenditure, what we observe is that reserves got totally wiped out, in the fixed ER closure, up to the point that they turn very negative. In the flexible ER closure, the depreciation is a little bit greater than the depreciation of the dollar in the respective simulation.