Gilberto Tadeu Lima and Gabriel Porcile

Economic growth and income distribution with heterogeneous preferences on the real exchange rate
Abstract: We develop a dynamic model of capacity utilization and growth taking into account the codetermination of international competitiveness (measured by the real exchange rate) and income distribution. It follows that distribution, capacity utilization, and growth vary with the real exchange rate depending on the source of change in the latter. Meanwhile, the nominal exchange rate (markup) varies whenever the real exchange rate differs from the level preferred by the government (capitalists). While there is a medium-run equilibrium when capitalists and the government share a preferred real exchange rate, convergence to it is not ensured. Indeed, when the government is primarily concerned with preserving workers share in income when managing the nominal exchange rate, a limit cycle arises: the economy goes through endogenous cyclical fluctuations in real exchange rate and growth that resemble the experience of several developing countries. Hence, growth regressions featuring the real exchange rate have to properly take into account the codetermination of the real exchange rate and the functional distribution of income. Key words: growth, distribution, capacity utilization, real exchange rate, markup. JEL classifications: F43, O11, O15

Gilberto Tadeu Lima is a full professor in the Department of Economics at the University of So Paulo, Brazil. Gabriel Porcile is a researcher in the Division of Production, Productivity and Management, Economic Commission for Latin America and the Caribbean (ECLAC), Santiago, Chile, and a professor in the Department of Economicsat the Federal University of Paran, Curitiba, Brazil. An earlier version of this paper was presented at the Thirty-Seventh Eastern Economic Association AnnualConference, New York, February 2527, 2011. The authors are grateful to the conferenceparticipants, in particular to Carlos Ibarra, and also to Arslan Razmi, Luiz Carlos Bresser-Pereira, Rafael Saulo Marques Ribeiro, and an anonymous referee for helpful comments. Any remaining errors are our own, however, and the opinions we express here do not necessarily reflect those of our institutions. Gilberto Tadeu Lima carried out most of his share of the work on this paper while visiting the Economics Department at the University of Massachusetts Amherst, whose hospitality and provisionof a stimulating academic environment are gratefully acknowledged.
Journal of Post Keynesian Economics/Summer 2013, Vol. 35, No. 4651 2013 M.E. Sharpe, Inc. All rights reserved. Permissions: www.copyright.com ISSN 01603477 (print)/ISSN 15577821 (online) DOI: 10.2753/PKE0160-3477350407

652 JOURNAL OF POST KEYNESIAN ECONOMICS

The growth-enhancing properties of an undervalued real exchange rate have recently been emphasized by a burgeoning literature (see, e.g., BresserPereira, 2010; Razmi et al., 2012; Rodrik, 2008). Meanwhile, income distribution and the real exchange rate are functionally related: whether the real exchange rate and the wage share, for instance, move in the same or opposite directions depends on the ultimate source of the change in either of them (Blecker, 1999). In fact, whether the domestic pricing equation includes imported intermediates also matters for the relationship between distribution and the real exchange rate, as intermediates represent another direct claimant on domestic income. Hence, whether exchange depreciations cause a rise in capacity utilization and growth also depends on the effect on aggregate demand of the accompanying change in distribution. The recent literature has dealt with the issue of the sustainability of a competitive real exchange rate (given some of its self-undermining externalities, such as an upward pressure on nontradable prices and/or nominal wages) and/or the related issue of avoiding negative effective demand externalities caused by an accompanying change in distribution. For instance, Razmi et al. (2012) assume unemployment in a two-sector model to show that changes in the real exchange rate affect employment in a way that allows the real exchange rate to be used to facilitate sustained capital accumulation. Porcile and Lima (2010) show how the elasticity of labor supply and endogenous (VerdoornKaldor) technological change play a role in preventing the real exchange rate from appreciating as the economy grows. Razmi (2010) shows that production diversification and learning by accumulation can avoid the negative distributional implications of exchange devaluations. Meanwhile, Rapetti (2011) shows that exchange depreciations will more likely trigger an acceleration of growth if they are implemented simultaneously with domestic demand management policies that prevent price inflation of nontradables and wage management policies that coordinate wage increases with tradable productivity growth. However, the literature has ignored inter- and intraclass conflicts over the preferred real exchange rate, with the latter being invariably (even if implicitly) conceived of as a consensual level, while the (often heated) public debates on exchange rate policy clearly illustrate that those conflicts are pervasive.1 As
1Using data from a World Bank survey administered to over 10,000 firms in 80 countries in 1999, Broz et al. (2008) offer evidence on a systematic pattern: tradables producers are more likely to be unhappy following an appreciation of the real exchange rate than are firms in nontradable sectors. The authors recall some dramatic events in the history of currency policy that involve conflicting preferences of social groups. For instance, the battle over the gold standard in the late nineteenth and early twentieth centurieswhether in the United States in the 1890s, the United Kingdom in 1925, or Latin America throughoutwas largely about which groups and sectors of the economy were likely to win and which to lose from being on that standard.

economic growth and income distribution 653

a result, it is reasonable to treat capitalists and the government as having preferred real exchange rates, which may differ from each other, and an innovation of this paper lies in the incorporation of this conflict of interests. In fact, the literature has not only ignored that the government, capitalists, and workers have different preferences in regard to the real exchange rate, but also has not fully considered that those players have different instruments to influence the course of the real exchange rate. Capitalists, for instance, taking the nominal wage, the nominal exchange rate, and some technological parameters as given, have a preferred real exchange rate to which a desired markup (their control variable) corresponds. Meanwhile, workers, taking the markup, the nominal exchange rate, and some technological parameters as given, have a preferred real exchange rate to which a desired nominal wage (their control variable) corresponds. Finally, the government, taking the markup, the nominal wage, and some technological parameters as given, has a preferred real exchange rate to which a desired nominal exchange rate (its control variable) corresponds. In a world in which the degree of openness has tended to increase steadily in the past twenty years and in which foreign competition is fierce, it is reasonable to assume that firms are well aware of the implications of the real exchange rate for market shares.2 Meanwhile, the real exchange rate preferred by the government usually has to balance two conflicting objectives: (1) when it is necessary to stabilize prices, an overvalued exchange rate is preferred, stimulating imports and curbing inflation; and conversely, (2) when the economy faces a severe disequilibrium on the external front, related to mounting trade deficits and/or an increasing cost of financing the external debt, an undervalued real exchange rate is preferred, despite its inflationary effects and worsening of distribution. In fact, many developing economies went through cycles in which overvaluation during stabilization programs was followed by the quest for competitiveness when external crises set in.3
In more closed economies, firms were able to set markups based principally on their desired share in income. In increasingly open economies, meanwhile, a key variable driving growth is the real exchange rate, which therefore cannot be ignored in pricing decisions. 3 For instance, currency overvaluation during stabilization programs occurred in Argentina and Chile in the 1970s, and in Brazil and Argentina in the 1990s. Fixed nominal exchange rates, controlled currency depreciations at a rate below the actual levels of domestic inflation, and high interest rates to achieve inflation targets, have been used in these countries as nominal anchors, leading in all cases to real exchange overvaluation (Ffrench-Davis and Ocampo, 2001). Meanwhile, several developing and East European economies in the 1980s had to keep the real exchange rate at a very high level (with disastrous implications, in most cases, for inflation and distribution) in order to be able to service the external debt.
2

654 JOURNAL OF POST KEYNESIAN ECONOMICS

As a result, given that (1) whether the wage share and the real exchange rate move in the same or opposite direction depends on the ultimate source of the change in either of them and that (2) capitalists, workers, and the government not only have heterogeneous preferences on the real exchange rate but also have different instruments under their control to influence its course, the main state variables in the dynamic analysis of this paper will be the markup and the nominal exchange rate rather than the wage share and the real exchange rate itself. Motivated by these considerations, we present a dynamic model of capacity utilization and growth in which the joint determination of distribution and international competitiveness (measured by the real exchange rate) is taken into account. In a given short run, how a change in the real exchange rate affects capacity utilization and growth depends on whether it has come about through a change in either the nominal exchange rate or the markup (or both). The reason is that a change in the real exchange rate (wage share) resulting from a change in either the nominal exchange rate or the markup does not leave the wage share (real exchange rate) unchanged. Meanwhile, a change in the profit share (real exchange rate) brought about by a change in the markup does not leave the real exchange rate (profit share) unchanged. Indeed, the rationale for our specification in terms of the markup and the nominal exchange rate (and not the distributive shares and the real exchange rate themselves) as state variables is the need to consider the codetermination of distribution and international competitiveness. Over the medium run, the nominal exchange rate (markup) varies in response to discrepancies between the actual real exchange rate and the real exchange rate preferred by the government (capitalists). But, as a change in the real exchange rate may result from a change in either the nominal exchange rate or the markup (or both), different specifications of the adjustment dynamics of these variables have distinct implications in terms of the existence and stability of a medium-run equilibrium solution for the real exchange rate, and thereby for the dynamics of capacity utilization and growth over the medium run. While there is a mediumrun (Nash) equilibrium characterized by the government and capitalists coming to share a preferred real exchange rate, whether the economy converges to it depends crucially on the specifics of the adjustment dynamics of the nominal exchange rate and the markup. In fact, when the government is primarily concerned with distribution in favor of workers when managing the nominal exchange rate, a limit cycle is obtained. In this specification, the economy experiences endogenous cyclical fluc-

economic growth and income distribution 655

tuations of its international competitiveness: capacity utilization and growth closely resembling the experience of several developing countries. Therefore, one empirical implication of the model is that growth regressions featuring the real exchange rate would have to properly take into account the codetermination of the real exchange rate and the functional distribution of income (or a proxy of the latter such as the real unit labor costs), which has not been the case in the literature. Alternatively, some procedure could be used to try to decompose a change in the real exchange rate into changes in the major components of the latter, namely, nominal exchange rate, nominal average (labor and, possibly, intermediate) costs, and profit margins (which has not been the case, either). Structure of the model The domestic economy is open to international trade but its government sector is omitted for simplicity (except for the consideration of a monetary authority managing the nominal exchange rate). Domestic production consists of a single good that can be used for both investment and consumption. Output is produced through a (fixed-coefficient) technical combination of two domestic (and homogeneous) factors of production, labor and capital, and an imported intermediate input. Capitalist firms in oligopolistic markets carry out domestic production. They produce (and hire domestic labor) according to effective demand, it being considered the only case in which excess capital capacity prevails in the short and medium runs. As a result, domestic labor employment is determined by domestic production: L = aX (1) where L is the domestic employment level, X is the domestic real output (hereafter, income or output) level, and a stands for the corresponding laboroutput ratio. The last, along with the other technical coefficients (domestic capital and imported intermediate input), is given at a point in time. Over the medium run, however, while the other technical coefficients are assumed to remain constant, the domestic laboroutput ratio is assumed to fall at a rate h due to purely labor-augmenting (Harrodneutral) technical change. The domestic price level (hereafter, price level) is given at a point in time, it being determined as a markup over average prime costs in the following way: P = z(aW + bePm) (2)

656 JOURNAL OF POST KEYNESIAN ECONOMICS

where z > 1 is the domestic markup factor (1 plus the domestic markup, hereafter markup), W is the domestic nominal wage, b is the imported intermediate input coefficient, e is the nominal exchange rate (home currency price of foreign currency), and Pm is the exogenously given foreign currency price of the imported intermediate input, which (for simplicity) is assumed to be equal to the international price level, P*. Given our focus, b and P* are assumed to remain constant over the medium run, so we simplify matters by normalizing both b and P* to unity. For the same reason, we further assume that the domestic unit nominal labor cost, aW, remains constant over the medium run, with the nominal wage being contractually defined to grow at the same rate as labor productivity. As we normalize the constant value of aW to unity, the price level defined by (1) simplifies to: P = z(1 + e) (3) The domestic economy is inhabited by two classes, capitalists and workers. Following the tradition of Kalecki (1971), we assume that these classes have different consumption behaviors. Workers provide labor and earn only wage income, all spent in consumption. Workers are always in excess supply, with the labor supply growing at a rate denoted by n. Capitalists receive profit income, the entire surplus over wages and imported inputs, saving all of it. Along with capitalists and workers, then, there is an external claimant on income represented by the imported intermediates. From (1) and (3), the share of labor in income, , is given by:
= Va = P 1 = 1 , z (1 + e)

(4)

where V = (W/P) is the domestic real wage (hereafter, real wage). As z > 1 and e > 0, the wage share as defined by (4) is automatically constrained to its economically relevant domain given by 0 < < 1. Hence, the real wage and the wage share vary negatively with the markup and the home currency price of foreign currency. The profit share, , can be expressed as:
= 1 WL beP* = 1 q, PX P

(5)

where q = eP1 is the real exchange rate (relative price of foreign goods), as explored below (as b and P* were normalized to unity, the share of imported intermediates in income is equal to the real exchange rate). As z > 1 and e > 0, the real exchange rate is automatically constrained to the economically relevant domain of the share of imported intermediates in income given by 0 < q < 1. As a rise in the markup lowers both the wage share and the real exchange rate, it unambiguously results in a rise in

economic growth and income distribution 657

the profit share. A rise in the nominal exchange rate (home currency per unit of foreign currency), however, lowers the wage share and raises the real exchange rate. It can be easily checked that (/e) + (q/e) = 0, so that a change in the nominal exchange rate leaves the profit share (though not necessarily the rateplease see below) unchanged. In fact, we can use (3) and (4) to rewrite (5) as follows: = 1 z1, (6) which confirms that the profit share varies positively with the markup but is not affected by a change in the nominal exchange rate. Besides, as z > 1, the profit share is automatically constrained to its economically relevant domain given by 0 < < 1. Interestingly, given the purpose of this paper, the preceding analysis reveals that it is not possible for both the wage and the profit shares to rise simultaneously (which would require, per (5), a fall in the share of imported intermediates in income). Indeed, while a change in the markup or the home currency price of foreign currency varies the wage share in the opposite direction, the same change either varies the profit share in the same direction or leaves it unaffected, respectively. With regard to the profit rate, the effect of a change in either the markup or the nominal exchange is ambiguous. To see this, note that the profit rate is given by: r = (1 q)u = (1 z1)u, (7) where r is the domestic profit rate (hereafter, profit rate), which is the flow of money profits divided by the value of the domestic capital stock, K, at output price, while u is domestic (capital) capacity utilization (hereafter, capacity utilization). As a result, although a higher markup raises the profit share, an accompanying change in capacity utilization may cause a net decrease in the profit rate. As the real exchange rate depends on the markup and the cost of foreign currency, it is then given in the short run. As we define the real exchange rate as the relative price of foreign goods, a higher q means a real depreciation of the domestic currency or improved price competitiveness of domestic goods.4 The real exchange rate can be rewritten in terms of its ultimate determinants as: e . q= (8) z (1 + e)
4Clearly, this specification implicitly assumes that domestic and foreign goods are imperfect substitutes, and that exported goods are qualitatively the same as domestic goods.

658 JOURNAL OF POST KEYNESIAN ECONOMICS

As a result, the real exchange rate varies positively (negatively) with the home currency price of foreign currency (markup), /e > 0 (q/z < 0), so that a change in the nominal exchange rate leads to a change in the same direction, but less than proportional, in the price level. As seen above, however, the real wage and the wage share in income vary negatively with the markup and the home currency per unit of foreign currency. The wage share and the real exchange rate are therefore functionally related in the following way:
( z, e ) = q( z, e ) . e

(9)

Hence, whether the wage share and the real exchange rate move in the same direction depends on what determinant changes. While a rise (fall) in the markup lowers (raises) the wage share and the real exchange rate, a nominal depreciation of the home currency (rise in the nominal exchange rate) lowers the wage share and raises the real exchange rate. As it turns out, how a change in the real exchange rate affects capacity utilization and growth in the short and medium runs depends on the accompanying change in the wage share. The reason is that the resulting change in aggregate demand depends on how the wage share and the share of imported intermediates are affected by a change in any of their common determinants. Given our assumption that workers income is all spent in consumption, while capitalists save all of their profit income, the ratio of aggregate consumption to the capital stock, gc, using (4), can be written as:
g c = u = u . z (1 + e)

(10)

Therefore, given capacity utilization, a higher markup or home currency price of foreign currency, by lowering workers share, leads to a fall in aggregate consumption as a proportion of the capital stock. Meanwhile, firms invest as follows: gi = gi(, , u, q), (11) where gi is investment as a proportion of the capital stock, > 0 dei , gi , gi > 0. We follow Marglin notes autonomous investment, and g u and Bhaduri (1990) in making investment depend positively on the profit share. A rationale for this specification is that, given capacity utilization, the current profit share is an index of expected future earnings; it both provides internal funding for investment and makes it easier to raise external funding. We follow Rowthorn (1981) and Dutt (1984) in

economic growth and income distribution 659

assuming that capital accumulation plans depend positively on capacity utilization due to accelerator-type effects. Now, even though we assume that all capital goods are domestically produced, there are reasonable theoretical and empirical grounds for including the real exchange rate as an argument in the investment function. In fact, a change in the cost of foreign currency has an effect on the price competitiveness of firms (both at home, given the competition of imported substitutes, and abroad, due to the change in export prices) and on the cost of imported inputs. Although the theoretical literature is not conclusive as to which of those effects is dominant, the empirical evidence seems to be more favorable to the assumption that investment i > 0. varies positively with the real exchange rate, so we assume gq Blecker (2007) investigates the effects of the real value of the U.S. dollar on aggregate investment in the U.S. domestic manufacturing sector, using data for 19732004. The main result of the estimation is a negative effect that is much larger than those found in previous studies, and counterfactual simulations show that U.S. manufacturing investment would have been 61 percent higher in 2004 if the dollar had not appreciated after 1995. Rapetti et al. (2012) use panel data for 184 countries with five-year time periods spanning 19602004, and find that real exchange rate undervaluations are (statistically) significant drivers of investment growth mainly in developing countries, this result being robust to different specifications, controls, and econometric methods. Bahmani-Oskooee and Hajilee (2010) investigate the effect of home currency depreciation on domestic investment using a time-series model of 50 countries for 19752006. However, while they find significant positive short-run effects of currency depreciation on investment in 43 out of the 50 countries, it is only in 21 countries that there are also long-run effects. Note, however, that the determinants of investment that are taken as given in the short run ( and q) depend on the state variables z and e, with varying positively with z, and q varying positively with e and negatively with z. Hence, while gi varies positively with e, it varies either positively or negatively with z. Let us then rewrite (11) as: gi = gi(z, e, u, ), (12)
i > 0. As for the sign of gi , it is positive (negative) if an increase in with ge z z, by raising , leads to a rise in investment that more (less) than offsets the concomitant fall in investment generated by the accompanying fall i > 0 (gi < 0) as a case of -effect (q-effect) dominance in q. We refer to gz z in investment behavior.

660 JOURNAL OF POST KEYNESIAN ECONOMICS

Finally, net exports (in real terms) as a proportion of the capital stock, g f, are given by: g f = g f(q, u, u*), (13) where u* is foreign capacity utilization, which is exogenously given. We assume that net exports vary positively (negatively) with the real exf > 0 (the change rate and foreign (domestic) capacity utilization, so that gq f f MarshallLerner condition holds), gu* > 0 and gu < 0. Now, the domestic determinant of net exports that is taken as given in the short run (q) depends on z and e, varying negatively with the former and positively with the latter. Let us then rewrite (13) as: It then follows that
f ge

g f = g f(z, e, u, u*). > 0 and

f gz

(14)

< 0.

The behavior of the model in the short run The short run is defined as a time frame in which the following variables can be taken as given: the stock of capital, K, the supply of labor, N, the markup, z, the laboroutput ratio, a, the money wage rate, W, the imported intermediate input coefficient, b, the nominal exchange rate, e (and hence the price level, P, the real exchange rate, q, and the distributive shares, and ), and the relevant foreign variables, u*, Pm, and P*. As domestic firms hold excess capacity and produce according to aggregate effective demand, the macroeconomic equality between leakages and injections is reached through changes in capacity utilization. This goods market equilibrium condition can be expressed as: u = gd gc + gi + g f, (15) where gd is aggregate effective demand as a proportion of the capital stock. As fiscal policy is omitted for simplicity, the domestic economys saving (gs = u gc) is spent on financing investment (gi) and the trade surplus (g f). As a result, if net exports are negative (g f < 0), the excess of domestic investment over national saving is financed by a corresponding inflow of foreign saving.5 Meanwhile, when there is a trade surplus or
Note that gs = u gc = ((z) + q(z, e)), so that national saving per unit of capital, gs, is also composed of saving supplemented by imports. Indeed, recall that q = [e/z(1 + e)], so that the share of imported inputs in income (which is equal to the real exchange rate) is increasing in the share of imports in variable cost (given by be/(aW + be) = e/(1 + e). As a higher import cost acts a vehicle for foreign saving, national saving per unit of capital therefore rises with the share of imports in variable cost.
5

economic growth and income distribution 661

deficit (g f 0) and capital is assumed not to depreciate, the growth rate of the capital stock (which is the growth rate of this one-good economy) is given by g = gi = gs g f. Substituting (10), (12), and (14) into (15) yields the following implicit solution for the short-run equilibrium value of domestic capacity utilization: u = gd = gc(z,e,u) + gi(z,e,u,) + g f(z,e,u,u*). (16) Given the demand-driven nature of the equilibrium capacity utilization, the corresponding stability condition is that u varies positively (negatively) when there is positive (negative) excess demand (given by EDG = gd u). This short-run stability condition is:
EDG c i = gu + gu + guf 1 < 0, u

(17)

i f which can be expressed alternatively as gc u + gu + g u < 1. Intuitively, in the neighborhood of the equilibrium capacity utilization, effective demand injections (investment and exports) should respond less than demand leakages (national savings plus imports) to a change in capacity utilization. Though we cannot obtain an explicit solution for capacity utilization (denoted by u), we can perform comparative statics exercises with respect to distribution and the real exchange rate. Note that we cannot hold the shares of wages and profits constant when the real exchange rate changes and vice versa, as shown by (4), (6), and (8). Indeed, a major motivation underlying our specification in terms of the markup and the nominal exchange rate (and not the distributive shares and the real exchange rate) is precisely the need to take into explicit consideration the codetermination of distribution and the real exchange rate. Differentiating (16) with respect to capacity utilization and the home currency per unit of foreign currency (and hence holding the markup constant), we obtain:

c i + ge + g ef u g e = , e

(18)

i f where D 1 gc u gu g u > 0 is the short-run stability condition given by (17). Therefore, for a nominal exchange depreciation (and, given that imports are only part of the total input costs, a real exchange depreciation) to raise equilibrium capacity utilization, the numerator of the above expression has to be positive. Now, while a nominal depreciation of

662 JOURNAL OF POST KEYNESIAN ECONOMICS

the home currency leading to a real depreciation causes a rise in both investment and net exports, it also causes, by reducing the wage share, a fall in consumption (recall that a nominal currency depreciation, despite leaving the profit share unchanged, makes for a rise in the share of intermediate imports in income, which is equal to the real exchange rate, at the expense of the wage share). Hence, nominal exchange depreciations raise (lower) equilibrium capacity utilization if what we term its price competitiveness effect (increase of the real exchange rate) is stronger than what we term its distributive effect (decrease of the wage share). Let us now differentiate (16) with respect to capacity utilization and the markup (and hence hold the nominal exchange rate constant) to obtain: c i + gz + g zf u' g z = , (19) z where D is as before. Hence, for a rise in the markup, which lowers the real exchange rate (and hence the share of imported intermediates in income) and the wage share, to raise equilibrium capacity utilization, the numerator of the above expression has to be positive. Now, a rise in the markup, by leading to a real exchange appreciation, causes a fall in both investment and net exports. Meanwhile, although a rise in the markup, by lowering the wage share and the share of imported intermediates in income, causes a fall in consumption, it also causes, by increasing the profit share, a rise in investment. As noted in the discussion following i is positive (negative) if an increase in the markup, by (12), the sign of gz raising the profit share, causes a rise in investment that more (less) than offsets the concomitant fall in investment generated by the accompanying fall in the real exchange rate. Therefore, if there is q-effect dominance in i < 0), as we termed it in the discussion following (12), a investment (gz rise in the markup unambiguously causes a fall in equilibrium capacity i > 0) which utilization. Meanwhile, a -effect dominance in investment (gz is strong enough to more than offset the accompanying fall in consumption and net exports following a rise in the markup causes a resulting rise < 0) as a case in equilibrium capacity utilization. We refer to uz > 0 (uz of -effect (q-effect) dominance in capacity utilization. Hence, while a necessary condition for a -effect dominance in capacity utilization i > 0), a q-effect domi(uz > 0) is a -effect dominance in investment (gz i nance in investment (gz < 0) causes an unambiguously q-effect dominance in capacity utilization (uz < 0). Our model shows that what is dubbed wage-led capacity utilization, or (more often) stagnationism (u > 0) in the post Keynesian literature more

economic growth and income distribution 663

broadly, can be brought about in an open economy by a fall in either the markup or the home currency price of foreign currency.6 As seen above, however, while a nominal appreciation of the home currency leading to a real exchange appreciation causes a fall in both investment and net exports, it also causes, by raising the wage share, a rise in consumption (recall that a nominal currency appreciation, despite leaving the profit share unchanged, causes a fall in the share of intermediate imports in income, which translates into a rise in the wage share). Therefore, wageled capacity utilization through a nominal currency appreciation obtains when what we termed as its distributive effect (raising of the wage share) is stronger than what we termed as its price competitiveness effect (real currency appreciation); in this case, a nominal currency appreciation causes a rise in consumption that more than compensates the accompanying fall in investment and net exports. Meanwhile, as also seen above, a fall in the markup, by leading to a real exchange depreciation, causes a rise in both investment and net exports. However, although a fall in the markup, by raising the wage share, causes a rise in consumption, it also causes, by reducing the profit share, a fall in investment. Therefore, wage-led capacity utilization through a fall in the markup obtains when there is what we termed q-effect dominance in capacity utilization; in this case, a fall in the markup causes a rise in consumption and net exports that more than compensates any eventual accompanying fall in investment. Meanwhile, what is dubbed profit-led capacity utilization or (more often) exhilarationism (u > 0) in the post Keynesian literature can be brought about, even in an open economy, solely by a rise in the markup, as the profit share does not depend on the home currency price of the foreign currency. Indeed, it is only the distribution between workers and the external claimant represented by the share of imported intermediates that is affected by changes in the nominal exchange rate. Now, as seen above, a rise in the markup, by leading to a real exchange appreciation, causes a fall in both investment and net exports. Meanwhile, although a rise in the markup, by lowering the wage share and the share of imported intermediates, causes a fall in consumption, it also causes, by increasing the profit share, a rise in investment. As noted in the discussion i is positive (negative) if an increase in the following (12), the sign of g z markup, by raising the profit share, causes a rise in investment that more (less) than offsets the concomitant fall in investment generated by the
6 In the ensuing discussion of distributional effects on capacity utilization and economic growth, we use a slightly modified version of alternative concepts compiled and clarified in Blecker (2002).

664 JOURNAL OF POST KEYNESIAN ECONOMICS

accompanying fall in the real exchange rate. Hence, profit-led capacity utilization through a rise in the markup obtains when there is what we termed -effect dominance in capacity utilization; in this case, a rise in the markup causes a rise in investment that more than compensates the accompanying fall in consumption and net exports. We cannot obtain an explicit solution for the growth rate either, but we can likewise perform comparative statics exercises with respect to distribution and the (nominal and real) exchange rate. Differentiating (12) with respect to the nominal exchange rate (and hence holding the markup constant) and using (18), we obtain:
i c i c guf ) + gu + g ef ) (1 gu ( ge g' g e (20) = , e where D is as before, so that the first term in parentheses in the numerator is positive. Now, giu, which measures the accelerator effect in the investment function, gie, which measures the impact of a nominal currency depreciation (which leads to a real currency depreciation) on investment, and gfe, which measures the effect of a nominal currency depreciation (which leads to a real currency depreciation) on net exports, are all positive. It is intuitive, then, that it takes a very strong (negative) effect of a nominal currency depreciation on consumption, given by gc e, to prevent such depreciation from resulting in a higher equilibrium growth rate. Hence, what is dubbed wage-led growth (g > 0) in the post Keynesian literature can be brought about, in an open economy, by a nominal exchange appreciation. But, it requires that the positive effect of a nominal currency appreciation (leading to a real currency appreciation) on consumption is strong enough to more than compensate the accompanying fall in investment and net exports. Let us now differentiate Equation (12) with respect to the markup (and hence hold the nominal exchange rate constant) and use Equation (19) to obtain:

i c i c guf ) + gu + g zf ) (1 gu (gz g' g z = , z

(21)

where D is again as before, so that the first term in parentheses in the i is ambiguous, it benumerator is positive. As seen above, the sign of gz ing positive (negative) if an increase in z, by raising , causes a rise in investment that more (less) than offsets the concomitant fall in investment i > 0 (gi < 0) generated by the accompanying fall in q. We referred to gz z above as a case of -effect (q-effect) dominance in investment behavior. Meanwhile, gc z, which measures the effect of a change in the markup

economic growth and income distribution 665

f , which measures the effect of a change in the on consumption, and g z i , which markup on net exports, are both negative. Therefore, given that gu measures the accelerator effect in the investment function, is positive, it follows that a situation of q-effect dominance in investment behavior i < 0) ensures that the equilibrium growth rate unambiguously varies (gz negatively with the markup; in this case, after all, a rise in the markup causes a fall in all components of aggregate effective demand. It is then the case that a necessary condition for the growth rate to vary positively i > 0) in investwith the markup is that there is -effect dominance (gz ment behavior. But (21) shows that it may take a strong positive effect of a rise in the markup on investment for the corresponding rise in the profit share to reduce a higher growth rate. Therefore, what is dubbed > 0) in the post Keynesian literature can be brought profit-led growth (g about, even in an open economy, solely by a rise in the markup; however, profit-led growth may require quite a strong positive response of investment to a rise in the markup. Finally, (21) also reveals the conditions under which what is dubbed wage-led growth (g > 0) is generated by a fall in the markup. Indeed, a fall in the markup, by leading to a real exchange depreciation, causes a rise in both investment and net exports. However, while a fall in the markup, by raising the wage share, causes a rise in consumption, it also causes, by reducing the profit share, a fall in investment. Hence, wageled growth through a fall in the markup unambiguously obtains when i < 0); in there is what we termed q-effect dominance in investment (gz this case, a fall in the markup causes a rise in all components of effective demand. Yet wage-led growth (g > 0) through a fall in the markup may still obtain even if there is what we termed -effect dominance in i < 0); it is necessary that the corresponding fall in investinvestment (gz ment is not strong enough to more than compensate the accompanying rise in consumption and net exports.

The behavior of the model in the medium run In the medium run we assume that the short-run equilibrium values of the endogenous variables are always attained, with the economy moving over time with changes in the following variables: K, whose dynamics is given by g above; z; e; N, whose growth rate is equal to n; a, which falls at an exogenous rate h; and W, whose growth rate is equal to h. Given our focus, the labor market (but not the labor supply) is assumed to play a passive role, which further justifies the assumption that the nominal wage grows at the same rate as labor productivity. We

666 JOURNAL OF POST KEYNESIAN ECONOMICS

then assume that the natural growth rate (given by gn = n + h) adjusts to g through changes in n, which implies (further assuming that such an adjustment is fast enough) a constant employment rate over the medium run. Therefore, we follow the behavior of the system over the medium run by formally examining the joint dynamic behavior of the markup and the home currency price of foreign currency. As shown by (3)(4) and (6)(8), steady-state values for z and e imply steady-state values for the price level and the distributive shares (including the share of intermediate imports, which is equal to the real exchange rate), and hence for the rates of capacity utilization and growth. The nominal exchange rate is managed by a crawling peg as follows: e = (qg q), (22) where e = (de/dt)(1/e) denotes the proportionate rate of change in the nominal exchange rate, qg is the real exchange rate preferred by the government, and > 0 is the speed of adjustment. While we follow Blecker (2011) in assuming that the government has a real exchange rate target, we do not take it to be exogenously given. The reaction function (22) is intended to convey the idea that the government, taking the markup as given, manages the cost of foreign currency so as to achieve its preferred real exchange rate. Given the markup, then, corresponding to a given real exchange rate preferred by the government, there is an exogenous preferred cost of foreign currency, eg. To make this intuition explicit, we use (8) to rewrite the above reaction function as:

eg e = e . z (1 + eg ) z (1 + e)

(23)

Meanwhile, capitalists manage the markup so as to achieve a preferred level of international competitiveness as measured by the real exchange rate: z = (q qc), (24) where z = (dz/dt)(1/z) denotes the proportionate rate of change of the markup, qc is the real exchange rate preferred by capitalists and > 0 is the speed of adjustment. Analogously to the reaction function governing the dynamics of the nominal exchange rate, the reaction function (24) is intended to convey the idea that capitalists, taking the cost of foreign currency as given, manage the markup so as to achieve their preferred

economic growth and income distribution 667

international competitiveness. Given the nominal exchange rate, then, corresponding to a given real exchange rate preferred by capitalists, there is an exogenous preferred markup, zc. To make this intuition explicit, we use (8) to rewrite the above reaction function as:

e e = z . z (1 + e) zc (1 + e)

(25)

We can solve for the medium-run equilibrium by imposing the conditions z = e = 0 on (25) and (23). As z and e are both positive, this yields the isoclines: and: e = eg. (27) It follows from (26) and (27) that z = zc and e = eg, respectively (where a prime sign again denotes an equilibrium value), and there is a unique medium-run equilibrium solution.7 Moreover, the corresponding equilibrium for the real exchange rate is given by q = qc = qg; the model endogenously produces an equilibrium real exchange rate stemming from evolving heterogeneous preferences over the exchange rate by the government and capitalists. Note that the equilibrium configuration so identified is stable, as the determinant and the trace of the Jacobian matrix corresponding to the system given by (23) and (25) (both evaluated at the nontrivial solution) are positive and negative, respectively. Graphically, the z = 0 isocline is parallel to the e axis, while the e = 0 isocline is parallel to the z axis, with the equilibrium E1 = (z, e) = (zc, eg) being a stable node (which implies that convergence of the cost of foreign currency to its medium-run equilibrium does not involve either over- or undershooting). As a result, not only is the government able to both set and achieve a desired nominal exchange rate, it is able to do so without thwarting the achievement of any desired markup by capitalists set independently of eg. Note that, by the same token, it is also possible for capitalists to set and pursue a desired markup without thwarting the achievement of any desired nominal exchange by the government set independently of zc.
7 In game-theoretic parlance, the reaction functions given by (23) and (25) can be interpreted as best-reply functions and the resulting medium-run equilibrium solution can therefore be defined as a Nash equilibrium.

z = zc

(26)

668 JOURNAL OF POST KEYNESIAN ECONOMICS

Let us specify the dynamic adjustment of the home currency price of foreign currency and the markup in the following alternative way (allowed by our decomposition of the real exchange rate into its determinants, the nominal exchange rate and the markup). Instead of (23) and (25) we have: and
e e = e z g (1 + e) z (1 + e)

(28)

e ec = z . z (1 + e) z (1 + ec )

(29)

We are assuming that e = eg and z = zc continuously, as if resulting from an instantaneous adjustment. Meanwhile, zg denotes the desired markup by the government (that is, given e, the markup implied by its preferred real exchange rate), and ec denotes the nominal exchange rate desired by capitalists (that is, given z, the nominal exchange rate implied by their preferred real exchange rate). We are using the fact that there are at least two reasons why, for instance, q is lower than qg: first, given z, we have e < eg; and second, given e, we have z > zg. Recall, however, that the government controls e rather than z. In the previous specification, the government, taking z as given, measures any gap between q and qg (and changes e accordingly) by the corresponding gap between e and eg. In this alternative specification, the government, taking e as given (as e = eg instantaneously), measures any discrepancy between q and qg (and changes e accordingly) by the corresponding gap between z and zg. And we are applying the same intuition to capitalists preferred real exchange rate. There are at least two reasons why, for instance, q is lower than qc: first, given e, we have z > zc; and second, given z, we have e < eg. Recall, however, that capitalists control z rather than e. In the previous specification, capitalists, taking e as given, measure any gap between q and qc (and change z accordingly) by the resulting gap between z and zc. In this alternative specification, capitalists, taking z as given (as z = zc instantaneously), measure any gap between q and qc (and change z accordingly) by the corresponding gap between e and ec. In the first specification, therefore, the lag is in the adjustment of e to eg, and of z to zc; in this second specification, the lag is in the adjustment of e to ec, and of z to zg.

economic growth and income distribution 669

One plausible rationale for the reaction function (28) is that the government aims to curb firms market power. Whenever policymakers consider the markup as too high, the government reduces e and puts more competitive pressure on domestic firms. In fact, such a policy may be part of an effort to reduce the price level. Conversely, if the government considers that the current level of z may compromise, for instance, firms internal funds for investment, it raises e (which amounts to an increase in the protection of domestic firms from foreign competition). Firms then react by reasserting their own perception of their market power, and will be less concerned with distribution and more concerned with not deviating much from what they see as a desirable international competitiveness. In any case, we can solve for the medium-run equilibrium by imposing the conditions z = e = 0 on (28) and (29), which yields e = ec and z = zg (where a prime sign denotes an equilibrium value), so that there is a unique medium-run equilibrium solution. Moreover, the corresponding equilibrium for the real exchange rate is given by q = qc = qg. Note that the equilibrium configuration so identified is saddle-point unstable, though, as the determinant of the Jacobian matrix corresponding to the system given by (28) and (29) (evaluated at the nontrivial solution) is negative, which is a necessary and sufficient condition for saddle-point instability to obtain. Graphically, the z = 0 isocline is parallel to the z axis, while the e = 0 isocline is parallel to the e axis, with the equilibrium configuration E1 = (z, e) = zg, ec) being saddle-point unstable. As a result, convergence of the nominal exchange rate to its medium-run equilibrium (which nonetheless takes place solely along the negatively sloped stable arm of the saddle path, assuming that e is in the vertical axis) does not involve either over- or undershooting. Alternatively, as the real exchange rate and the workers share are inversely related, what if the government is more concerned with distribution (or, more specifically, with the workers share) than international competitiveness when managing the cost of foreign currency? In the previous specification, (28) shows that when z rises, the government reacts by raising e. However, both changes affect the wage share negatively, so that workers my have enough influence over the government to persuade it to compensate for a rise in z with a reduction in e. Note that now it is the real exchange rate that is affected negatively by both changes, though the workers share may end up rising. Formally, this pro-labor crawling peg, means that the parameter in (28) becomes negative. As in the previous specification, it follows that e = ec and z = zg, respectively, so that there

670 JOURNAL OF POST KEYNESIAN ECONOMICS

Figure 1 Endogenous cyclical fluctuations of real exchange rate, income distribution, and economic growth

e =0

A ec E1 D C C B

z =0

zg

is a unique medium-run equilibrium. Also, the corresponding equilibrium for the real exchange rate is q = qc = qg. A limit cycle obtains, though, as the determinant and the trace of the Jacobian matrix corresponding to the system given by (28) and (29) (evaluated at the nontrivial solution) are now positive and zero, respectively. As Figure 1 shows, the z = 0 isocline is parallel to the z axis, while the e = 0 isocline is parallel to the e axis, which yields the equilibrium configuration E1 = (z, e) = zg, ec). In this specification, the government manages the nominal exchange rate to ensure that workers share (and the implied real exchange rate) is realized. As a result, although there is a (unique) stationary point, a limit cycle obtains. For any initial values of the nominal exchange rate and the markup, the solution of the system yields a closed trajectory: if left undisturbed, the model produces conservative cyclical fluctuations in the nominal exchange rate and the markup (and hence in the real exchange rate, distribution, capacity utilization, and growth). Consider point A in Figure 1. As z = zg but e > ec, it follows that q > q, so that the reaction functions make the economy move over time with rising markup and falling cost of foreign currency. In point B, e = e but z > zg, so that q < q. On its way from A to B the economy crosses point A, in which q = q. Now, what happens to distribution, capacity utilization, and growth as the economy travels from A to B depends on the signs of the corresponding comparative statics, as discussed in the previous section. Consider point C now. As z = zg but e < ec, it follows that q < q, so that the reaction

economic growth and income distribution 671

functions make the economy move over time with rising cost of foreign currency and falling markup. In point D, e = ec but z < zg, so that q > q. On its way from C to D the economy crosses again the isocline through the origin (point C) in which q = q. As q > q in D, the reaction functions make the economy move over time with rising markup and cost of foreign currency, which keeps q > q up to point A. From a policy perspective, one way of comparing the results derived above is as follows. If the economy is in a short run with the government and capitalists having different preferences on the real exchange, it is worth wondering (1) whether there is a medium-run equilibrium in which capitalists and the government come to have a common preferred real exchange rate, (2) whether the economy converges to such an equilibrium solution in which preferences on the real exchange rate are homogeneous, and (3) what the corresponding implications are in terms of the dynamics of capacity utilization and economic growth. While for issue (1) the answer is that there is indeed a medium-run equilibrium characterized by capitalists and the government coming to share a preferred real exchange rate, the answer to issues (2) and (3) depend on the specifics of the adjustment dynamics of the nominal exchange rate and the markup. The reason is that a deviation of the actual real exchange rate from some preferred level can be corrected through a change in either the nominal exchange rate or the markup (or both), as detailed in the preceding section. In the first specification, the government manages the nominal exchange rate, caring primarily about the international competitiveness of the economy, while capitalists adjust the markup caring about both their international competitiveness and share in domestic income. Capitalists and the government nonetheless properly manage the corresponding determinant of the real exchange rate in response to deviations of its current value from the value implied by the respective preferred real exchange rate. As a result, the economy converges (noncyclically) to the medium-run equilibrium characterized by capitalists and the government sharing a preferred real exchange rate. In the second specification, the government similarly manages the nominal exchange rate caring primarily about the international competitiveness of the economy, while capitalists similarly adjust the markup caring about both their international competitiveness and share in domestic income. However, capitalists and the government now manage the corresponding determinant of the real exchange rate in response to deviations of the current value of the other determinant from the value implied by the respective preferred real exchange rate. As it turns out, the economy converges to the medium-run equilibrium with capitalists and the government sharing a preferred real exchange

672 JOURNAL OF POST KEYNESIAN ECONOMICS

rate only by a happy coincidence (that is, only if it happens to be in the stable arm of the saddle path). In the third specification, meanwhile, the government manages the nominal exchange rate caring primarily about preserving the share of workers in domestic income, while capitalists again adjust the markup caring about both their international competitiveness and share in domestic income. As in the previous specification, capitalists and the government manage the corresponding determinant of the real exchange rate in response to deviations of the current value of the other determinant from the value implied by the respective preferred real exchange rate. It follows that a limit cycle obtains: the economy never converges to the medium-run equilibrium with capitalists and the government sharing a preferred real exchange rate from any initial point that is not the equilibrium itself. With regard to the implications of these alternative specifications for the medium-run dynamics of capacity utilization and economic growth, meanwhile, the comparative statics performed in the preceding section reveal that they cannot be easily and unambiguously ascertained. The reason is that whether international competitiveness and income distribution (measured by the wage share, for instance) move in the same or opposite direction depends on whether the (common) source of change is the nominal exchange rate or the markup. Besides, changes in the real exchange rate and income distribution affect the components of aggregate effective demand in different and sometimes opposite ways. Finally, the dynamic analysis performed in this section shows that whether the nominal exchange and the markup move in the same or opposite directions depends on whether (and from what initial position) the economy converges to or diverges from the medium-run equilibrium. Indeed, this paper contributes to the existing literature mostly by precisely mapping out all of these ambiguities. Summary Although a growing literature has emphasized the growth-enhancing properties of an undervalued real exchange rate, interclass conflicts over the preferred real exchange rate have been ignored. In fact, the literature has not only ignored the different preferences of government, capitalists, and workers in regard to the real exchange rate, but has also not fully considered that these players have different instruments to influence its course. Meanwhile, the real exchange rate and distribution are functionally related, and whether they move in the same or opposite

economic growth and income distribution 673

directions depends on the source of the change in either the exchange rate or distribution. This paper contributes to the literature by developing a dynamic model of capacity utilization and growth in which the codetermination of distribution and the real exchange rate (and resulting heterogeneous preferences in regard to the latter) is taken into account. In the short run, how a change in the real exchange rate affects capacity utilization and growth depends on whether it has resulted from a change in either the nominal exchange rate or the markup (or both). Over time, the nominal exchange rate (markup) varies in response to discrepancies between the actual real exchange rate and the real exchange rate preferred by the government (capitalists). But, as a change in the real exchange rate may result from a change in either the nominal exchange rate or the markup (or both), different specifications of the adjustment dynamics have differing implications in terms of the existence and stability of a medium-run equilibrium for the real exchange rate, and thereby for the dynamics of capacity utilization and growth. While there is a mediumrun equilibrium with capitalists and the government coming to share a preferred real exchange rate, convergence to it is not ensured. In fact, when the government is primarily concerned with preserving workers share in income when managing the nominal exchange rate, a limit cycle obtains and the economy goes through endogenous cyclical fluctuations in the real exchange rate and growth rate, resembling the experience of several developing countries. Hence, one empirical implication of the model is that growth regressions featuring the real exchange rate should properly take into account the codetermination of the real exchange rate and the functional distribution of income, which has not been the case in the corresponding literature. REFERENCES
Bahmani-Oskooee, M., and Hajilee, M. On the Relation Between Currency Depreciation and Domestic Investment. Journal of Post Keynesian Economics, 2010, 32 (4), 645660. Blecker, R.A. Kaleckian Macro Models for Open Economies. In J. Deprez and J. Harvey (eds.), Foundations of International Economics: Post Keynesian Perspectives. London: Routledge, 1999, pp. 116149. . Distribution, Demand and Growth in Neo-Kaleckian Macro-Models. In M. Setterfield (ed.), The Economics of Demand-Led Growth. Aldershot, UK: Edward Elgar, 2002, pp. 129152. . The Economic Consequences of Dollar Appreciation for U.S. Manufacturing Investment: A Time-Series Analysis. International Review of Applied Economics, 2007, 21 (4), 491517.

674 JOURNAL OF POST KEYNESIAN ECONOMICS

. Open Economy Models of Distribution and Growth. In E. Hein and E. Stockhammer (eds.), A Modern Guide to Keynesian Macroeconomics and Economic Policies. Cheltenham, UK: Edward Elgar, 2011, pp. 215239. Bresser-Pereira, L.C. Globalization and Competition. Cambridge: Cambridge University Press, 2010. Broz, J.L.; Frieden, J.; and Weymouth, S. Exchange Rate Policy Attitudes: Direct Evidence from Survey Data. IMF Staff Papers, 2008, 55 (3), 417444. Dutt, A.K. Stagnation, Income Distribution and Monopoly Power. Cambridge Journal of Economics, 1984, 8 (1), 2540. Ffrench-Davis, R., and Ocampo, J.A. The Globalization of Financial Volatility. In R. Ffrench-Davis (ed.), Financial Crises in Successful Emerging Economies. Washington, DC: Brookings Institution Press/ECLAC, 2001, pp. 137. Kalecki, M. Selected Essays on the Dynamics of the Capitalist Economy. Cambridge: Cambridge University Press, 1971. Marglin, S., and Bhaduri, A. Profit Squeeze and Keynesian Theory. In S.A. Marglin and J.B. Schor (eds.), The Golden Age of Capitalism. Oxford: Oxford University Press, 1990, pp. 153185. Porcile, G., and Lima, G.T. Real Exchange Rate and Elasticity of Labour Supply in a Balance-of-Payments Constrained Macrodynamics. Cambridge Journal of Economics, 2010, 34 (6), 10191039. Rapetti, M. Policy Coordination in a Competitive Real Exchange Rate Strategy for Development. Working Paper 2011-09, Department of Economics, University of Massachusetts Amherst, 2011. Rapetti, M.; Skott, P.; and Razmi, A. The Real Exchange Rate and Economic Growth: Are Developing Countries Different? International Review of Applied Economics, 2012, 26 (6), 735753. Razmi, A. The Exchange Rate, Diversification, and Distribution in a Modified Ricardian Model with a Continuum of Goods. Working Paper 2010-06, Department of Economics, University of Massachusetts Amherst, 2010. Razmi, A.; Rapetti, M.; and Skott, P. The Real Exchange Rate and Economic Development. Structural Change and Economic Dynamics, 2012, 23 (2), 151169. Rodrik, S. The Real Exchange Rate and Economic Growth. Brooking Papers on Economic Activity, 2008 (Fall), 365412. Rowthorn, B. Demand, Real Wages and Economic Growth. Thames Papers in Political Economy, 1981 (Autumn), 139.