Inequality, emulation and debt: The occurrence of different growth regimes in the age of financialization in a stock-flow consistent model

ABSTRACT

In the era of financialization, increasing income inequality could be observed in most developed and many developing countries. Despite similar developments in inequality, the growth performance and drivers for growth differed markedly among countries, allowing clusters of different growth regimes to be identified. Among them are 2 extreme types: the debt-led private-demand boom and the export-led mercantilist economies. Whereas the former rely mainly on credit-financed household consumption in order to compensate for the potential lack of demand (associated with the depressing effect of financialisation), the latter rely on net exports as the main driver of aggregate demand. Using a stock-flow consistent model it will be demonstrated how increasing inequality, depending on a countries institutional structure and regulatory framework, affects growth differently, explaining the occurrence of both regime types.

KEYWORDS:

• Debt
• emulation
• finance-dominate capitalism
• financialization
• international imbalances
• stock-flow consistent model

JEL CLASSIFICATIONS:

• E02
• E12
• E21
• E25
• E44
• E65
• F40
• F41
• F43

Acknowledgments

I would like to thank in particular Eckhard Hein and Hansjörg Herr who were always available for consultation and very fruitful discussions during the writing process. I’m also indebted to Paul Beckta, Dirk Ehnts, Marc Lavoie, Gennaro Zezza, and Thomas Theobald, who gave me helpful advices on the model. I started with SFC-modeling at the Winter School “Applied Macro-Modelling: Fully Scalable Models” at the University of Limerick and would like to thank the organizers Eugenio Caverzasi, Antoine Godin and Stephen Kinsella and the other participants. Earlier versions of the paper were presented at the FESSUD annual conference, October 16, 2015, in Lisbon; the 19th Conference of the Research Network Macroeconomics and Macroeconomic Policies, October 23, 2015; in Berlin and in the special lecture “Financialization and the Open Economy: Fresh Perspectives on Money and Credit,” October 26, 2015, at the Bard College in Berlin. I would like to thank the participants for very helpful comments and suggestions. Also, I would like to thank the two FESSUD commentators Wlodzimierz Dymarski and Gilad Isaacs, who gave me very helpful hints and advices. Finally, for editing assistance I am very grateful to Finn Cahill-Webb. Also, I would like to thank two anonymous referees for their reviews and their useful comments. Remaining errors are, of course, my own.

Notes

¹A country is described as wage-led when an increase in the wage share increases aggregate demand, or profit-led when an increase in the profit share increases aggregate demand (Bhaduri and Marglin1990). A range of empirical studies have shown that based on their domestic demand aggregates most countries are wage-led, even though for some countries the results can differ when net-exports are also considered (Bowles and Boyer 1995, Stockhammer and Ederer 2008, Stockhammer, Onaran, and Ederer 2009, Stockhammer, Hein, and Grafl 2011, Onaran and Galanis 2012).

²For research on different savings rates by different income groups, see Dynan, Skinner, and Zeldes (2004) or more recently Alvarez-Cuadrado and Vilalta (2012). See also Brown (2004) for discussion and simulations on the effect of income inequality on aggregate consumption in the United States.

³Some authors in more complex models have combined arguments of both types and shown how they interact and can lead to different macroeconomic dynamics, for example, Kumhof et al. (2012), Belabed, Theobald, and van Treeck (2017) and Cardaci and Saraceno (2016). We will have a closer look at some of them below.

⁴The entire set of model equations and a full list of all variables can be found in Appendix A.

⁵This compares to other specifications, for example, in a Harrodian or Marxian tradition, where distribution is the adjusting variable. For a good overview of different specifications and their implications, we recommend Hein (2017).

⁶This choice is justified by mainly two reasons: It keeps the model simple and it will make it easier to distinguish the effects of changes in different forms of inequality. In addition, there are many potential factors affecting distribution and not one straightforward option to select (see, e.g., the summary by Dutt 2012). Although not the focus of this article, this leaves room for interesting extensions of the model for future research: for example, as mentioned earlier, financialization does affect inequality through various channels and so an interesting pathway would be to make inequality dependent on financialization. Furthermore, feedback loops between employment levels, distribution, and growth, as for example shown by Goodwin (1972), may result in interesting macro-dynamics.

⁷Empirical evidence for this type of consumption function is provided by Kim, Setterfield, and Mei (2015) for U.S. households.

⁸This assumption is in line with the argument by Frank (2007) that consumption behavior is most heavily influence by reference groups close to one in rank, time, and space.

⁹Our consumption function is oriented along the same lines as the consumption function found in Belabed, Theobald, and van Treeck (2017). Other SFC models including a Veblenian component in their consumption function are, for example, Caverzasi and Godin (2015), which make worker household consumption dependent on consumption of rentier households (and the expected loan supply by banks) or Sawyer and Veronese Passarella (2017), which make loan demand of workers depend on the per capita consumption gap between workers and rentiers. For the discussion of consumption emulation effects in the literature, see, for example, Duesenberry (1949), Frank (2007), Frank, Levine, and Dijk (2014).

¹⁰These imply a range of adding up constraints, which ensure consistency of the chosen parameters for the portfolio decision of households. For an overview of those, see Godley and Lavoie (2007, pp. 141–146).

¹¹See, for example, Arghyrou and Chortareas (2008), European Commission (2010) and Carrasco and Peinado (2015) on Euro area member countries in general, and by Kollmann et al. (2015) and Storm and Naastepad (2015) on Germany, in particular.

¹²This clearly is an abstraction, because bank lending is only part of total international financial flows. However, we abstain here from introducing other assets to keep the model simple. Also, in the years before and during the recent financial crisis bank lending played an important role, especially in the Euro area. Prior to the financial crisis, banks from core countries like Germany and France have increased their lending to the later crisis countries Greece, Ireland, Italy, Portugal, and Spain substantially, to then withdraw it rapidly when the crisis hit (Foster, Vasardani, and Ca’ Zorzi 2011).

¹³This roughly equals the average world growth rate in the 2000s before the financial and economic crisis.

¹⁴See FESSUD Studies 18–34, available at http://fessud.eu/studies-in-financial-systems/.

¹⁵The values do not correspond to the values found in any particular country, but are close to what can be found in a typical western industrialized country. The shares in GDP would roughly fit the values found for the Euro area in the year 2000. However, the share of consumption is slightly lower and investment and government expenditure are slightly higher. In the baseline scenario both model economies do not clearly exhibit features of the export-led mercantilist or the debt-led private-demand types of development. This is not surprising, because in this model it is expected that inequality is the driver for the divergence of the macroeconomic results in the two cases and in the baseline scenarios we have assumed a fairly low degree of inequality. The specific macroeconomic features of the two models will only occur when inequality is increased.

¹⁶To investigate the effects of the different experiments we simulated both model economies until they converged to a steady state. We introduced the shock in Period 100 and ran the model for another 400 periods to obtain the new steady states. Each period should be considered as one year. With 500 periods our simulations cover a very long period of 500 years. This is due to the fact that after a shock the model is adjusting for a relatively long time towards the new steady state. However, as one can see in the figures, the most relevant deviations take place within 10 to 20 periods after a shock, whereas the changes thereafter are of relatively small magnitude.

¹⁷Here a key mechanism to drive the economy back to its initial growth path can be observed: Given the lower propensity to consume of the Worker 2 households, they save more of their income. This slows down domestic growth, but the rest of the world, as we assume, remains unaffected. This improves the trade balance of the domestic economy and domestic wealth-to-income ratios increase, which slowly lifts growth, because of the wealth term in the consumption function, until it has returned to a new steady state. However, although the growth rate returns to its initial level, there can be substantial changes in the level and the demand components of GDP. This illustrates a more general point: adjustment mechanism of SFC models toward the steady state are based on reaction functions to disequilibria. All sectors set themselves norms or targets and act in line with these and the expectations they form about the future. If expectations do not realize, there will be the depletion or piling up of stocks, which act as signals to change behavior. Therefore, a kind of autopilot is slowly adjusting behavior (Godley and Lavoie 2007). As described, in this model, for the household sectors adjustment occurs through the wealth term in the consumption function, whereas the government adjusts its expenditures according to its deficit and the firm sector is adapting its investment behavior to the utilization rate of the capital stock.

¹⁸See Cynamon and Fazzari (2008)/FESSUD studies on Financial Systems 1–17 for detailed studies on the spread of financial innovations that facilitated easier access to credit (http://fessud.eu/studies-in-financial-systems/).

¹⁹Here we equate higher financial fragility with higher debt-to-income ratios.

²⁰Recently, Stockhammer and Wildauer (2016) try to estimate the effect of expenditure cascades on consumption in a panel regression of 18 OECD countries covering the period 1980–2013. They do not find evidence for the validity of this explanation, but rather see the development of house prices as relevant factor. However, in their conclusions they acknowledge that it is possible that the relative income hypothesis does correctly describe desired consumption, but that it realizes only into actual consumption when due to the rising real estate prices credit constraints are lifted and therefore “effectively, rising property prices rather than inequality is identified as the main determinant of expenditure cascades.” Stockhammer and Wildauer 2016, p. 1628).

Additional information

Funding

This work was supported by the Seventh Framework Programme [Grant Number 266800].

Notes on contributors

Daniel Detzer

Daniel Detzer is affiliated with the Berlin School of Economics and Law, Institute for International Political Economy (IPE), Berlin, Germany.