Political aspects of household finance: debt, wage bargaining, and macroeconomic (in)stability

Abstract

The recent literature has shown that income inequality is one of the main causes of borrowing and debt accumulation by working households. This article explores the possibility that household indebtedness is an important cause of rising income inequality. If workers experience rising debt burdens, their cost of job loss may rise if they need labor-market income to continue borrowing and servicing existing debt. This, in turn, will reduce their bargaining power and increase income inequality, inducing workers to borrow more to maintain consumption standards, and so creating a vicious circle of rising inequality, job insecurity, and indebtedness. We believe that these dynamics may have contributed to observed simultaneous increases in income inequality and household debt prior to the recent financial crisis. To explore the two-way interaction between inequality and debt, we develop an employment rent framework that explicitly considers the impact of workers’ indebtedness on their perceived cost of job loss. This is embedded in a neo-Kaleckian macro model in which inequality spurs debt accumulation that contributes to household consumption spending and hence demand formation. Our analysis suggests that (a) workers’ borrowing behavior plays a crucial role in understanding the character of demand and growth regimes; (b) debt and workers’ borrowing behavior play an important role in the labor market by influencing workers’ bargaining power; and (c) through such channels, workers’ borrowing behavior can be a decisive factor in the determination of macroeconomic (in)stability.

Keywords:

• Consumer debt
• employment rent
• cost of job loss
• bargaining power
• income distribution
• growth
• stability

JEL classifications:

• E12
• E21
• E24
• E44
• O41

Notes

1 These observations dovetail with more general empirical findings that “financialization” has been a source of increased income inequality (Darcillon 2016).

2 Note that, together with their ability to bargain (as reflected in, for example, the structure of labor law), we hereby treat the very willingness of workers to bargain as one dimension of workers’ bargaining power. In what follows we refer to changes in the employment rent as affecting workers’ bargaining power on the understanding that it does so by affecting the willingness of workers to bargain, as described above.

3 We assume throughout that $D_W\hspace{0.17em}>\hspace{0.17em}0$, thereby excluding the possibility that working households might act as net creditors. This is consistent with the stylized facts of household balance sheets in recent decades (Barba and Pivetti 2009; Carr and Jayadev 2015).

4 Note that implicit in this statement of workers’ budget constraint are the assumptions that workers do not save, and that workers’ debts are not amortized but instead roll over from period to period. In other words, debt servicing consists only of meeting interest payable on outstanding debt, so that as they continue to borrow, the total outstanding indebtedness of working households increases over time. Again, this is consistent with the stylized facts of household balance sheets in recent decades (Barba and Pivetti 2009; Carr and Jayadev 2015).

5 For simplicity, the price of equity is fixed and normalized to one.

6 This consistency condition reflects that one agent’s expenditure must be equal to another agent’s income in the economy as a whole.

7 Investment behavior is an important and contested subject in Post Keynesian models. Since our focus is on workers’ borrowing and associated consumption behavior, we use a simple neo-Keynesian investment function and leave extensions of our analysis that involve changing the form of the of the investment function to future research.

8 Normalized, as are the other variables in our model, by the capital stock.

9 Equivalently, capitalists’ saving is given by $S_R=s_R(\Pi +i{D}_W)$. Recalling capitalists’ budget constraint from the second column of table 2, equilibrium in the consumer credit market therefore requires that: (10) $${\dot{D}}_W=s_R(\Pi +i{D}_W)-\dot{Q}$$(10)

Note that this equality holds in equilibrium, when saving by capitalist households fully funds autonomous spending by workers as well as investment by firms. It does not imply that our model is subject to a savings-in-advance constraint, which would prevent it from being demand-led. In other words, some form of endogenous credit creation is required in order for the model to traverse from one steady state configuration to another. Because our interest is purely in the characteristics of the steady state itself, however, we abstract from this monetary feature of disequilibrium adjustment for the sake of simplicity.

10 Note that the size of the wage share itself will also influence this possibility, and hence whether the demand regime is wage- or profit-led.

11 Note that with the neo-Keynesian investment function in (3), in the absence of worker borrowing and indebtedness ($f(d_W,\psi ), d_W=0$), we have: (14a) $$u=\frac{{\kappa }_0}{(s_R-{\kappa }_0)(1-\psi )}$$(14a) and: (17a) $$\frac{\partial u}{\partial \psi }=\frac{{\kappa }_0(s_R-{\kappa }_0)}{{\left[\right(s_R-{\kappa }_0\left)\right(1-\psi \left)\right]}^2}>0$$(17a) under standard assumptions about the Keynesian stability condition (s_R – κ₀ > 0). As these results demonstrate, worker borrowing and indebtedness are solely responsible for transforming an unambiguously wage-led demand regime into one that can be either wage- or profit-led.

12 Note that our results here are in line with previous studies that examine the implications of debt-financed consumption for wage-led vs. profit-led demand and growth regimes (Setterfield and Kim 2017).

13 This is assured if ${\kappa }_0\hspace{0.17em}>\hspace{0.17em}i({\kappa }_r\hspace{0.17em}-\hspace{0.17em}{s}_R)s_R$, which is a reasonable condition since $0\hspace{0.17em}<\hspace{0.17em}i,s_R\hspace{0.17em}<\hspace{0.17em}1$.

14 Note that the size of workers’ debt burdens will also influence this possibility, and hence whether the demand and growth regimes are debt-led or debt-burdened.

15 This is consistent with our treatment of $h$ as exogenously given (and hence independent of local adjustments in $u$ towards its steady state value). To see this, suppose we think of $h$ as being proxied by the current rate of unemployment $U$, so that: (24) $$h=U=1-\frac{L}{N}=1-\frac{L}{Y}\frac{Y}{K}\frac{K}{N}=1-auk$$(24) where $N$ is the labor force, $a$ is the ratio of the total number of persons employed to total output, and $k$ is the ratio of the capital stock to the labor force (not to be confused with the capital-labor ratio associated with the production technology on the supply side). Since by definition $\dot{k}=k(g_K\hspace{0.17em}-\hspace{0.17em}n)$, where $n$ is the rate of growth of the labor force, we can treat $k$ as constant ($\dot{k}=0$) by postulating that $n=g_K$ – that is, the rate of growth of the labor force adjusts endogenously to meet the needs of a growing capitalist economy through inter-sectoral and/or inter-regional migration of a global “reserve army” of labor. With $k$ thus fixed, any local variation in $u$ implied by our stability analysis is absorbed in our model by local variation in $a$, leaving $h$ unchanged.

16 Recall that, as previously explained, local variations in $u$ are assumed to be absorbed by local variations in hours worked (rather than the unemployment rate). It is, then, strictly the binary distinction between having or not having a job (rather than variation in hours worked by those employed) that affects the capacity to borrow and hence the value of the employment rent in our analysis. In fact, variation in hours worked, and the tightness or slack in the labor market that this might appear to imply, has no effect on the value of the employment rent in our model.

17 Recall that the transition from employment to unemployment is a credible threat not a physical state in this model. Hence setting the income associated with unemployment to zero does not mean that our debt dynamics are affected by default as a result of previously employed workers becoming unemployed (and earning no income), because no such transitions of state (from employment to unemployment) actually occur.

18 We assume that ${\dot{d}}_W$ and $\dot{\psi }$ are bounded and continuously differentiable functions that pass through the value of zero in the range of $d_W\in R^{+}$ and $\psi \in (0,1)$.