西方经济学评论 2014卷 第1辑 (总第4辑)
上QQ阅读APP看书,第一时间看更新

3.Minsky cycle as a Lotka-Volterra like system

Minsky himself did not formulate formal mathematical models that reflect his basic idea on the“financial instability hypothesis”.However, a lot of mathematical models that were inspired by Minsky's idea have been produced up to now since the seminal paper by Taylor and O'Connell(1985)was published.Following examples are some of such works. Arena and Raybaut(1998), Asada(2001,2004,2006), Bhaduri(2012), Charles(2008a,2008b), Charpe, Chiarella, Flaschel and Semmler(2011), Delli Gatti and Gallegati(1995), Foley(1987), Jarsulic(1990), Keen(2000), Kuroki(1994), Pally(1996)and papers in Semmler(ed.)(1989).Nasica(2000)provides an excellent survey of the related topics.In this section, we shall summarize the essence of the model that was presented by Asada(2001)as a typical example of such mathematical models.This approach interpret the“Minsky cycle”as a kind of Lotka-Volterra system of mathematical biology, which is based on the dynamic interaction of predator and prey.See Gandolfo(2009)chap.23 as for the exposition of the Lotka-Volterra system.Goodwin(1967)is very famous contribution as a quite interesting application of this system to macrodynamic economic theory.It must be noted, however, our formulation is the only partial formalization of Minsky's analytical scheme.For example, it is not clear whether the“subprime crisis”fits our Lotka-Volterra like model well.

The reduced form of a system of equations that was presented by Asada(2001)consists of the following two-dimensional system of differential equations.

where a dot over the symbol is time derivative, and the meanings of the symbols are as follows.

d=D/K=private debt-capital ratio.y=Y/K=income-capital ratio.D=stock of the real debt of the private firms.K=real capital stock that is owned by the private firms.Y=real national income.α=parameter that reflects the adjustment speed of the disequilibrium in the goods market.Variable y is used as a surrogate variable that reflects the degree of the utilization of capital stock and the rate of employment(1-rate of unemployment)in the labor market.This procedure, which was adopted by Franke and Asada(1994), can simplify the analysis by saving one state variable, the rate of employment(e).Although this procedure is not precise method but only an approximation, we can justify this kind of simplification due to high correlation between two variables y and e(see Franke and Asada,1994).In this simplified version of the model, the fixed price is assumed, and the government's economic activity and international transaction are abstracted from.Later, we shall interpret the extended models that can treat the price movement and macroeconomic stabilization policies by the government including the central bank.

Furthermore, each partial derivative in this model becomes as follows because of some economic reasons(cf.Asada 2001).Asada(2001)showed that we have a set of inequalities(2)if, gy, |gd|in Eq.(3)and the firms'marginal propensity of internal retention are sufficiently large.

Eq.(1)(i)can be derived from the fact that the firms'investment expenditure that exceeds the firms'internal finance must be debt-financed.It is assumed that the firms are debtors and the capitalists are creditors, and the issues of new shares are abstracted from.Eq.(1)(ii)expresses the“quantity adjustment”process in the goods market disequilibrium, which implies that the rate of utilization of the capital stock fluctuates according as the excess demand in the goods market per capital stock is positive or negative.In both equations(1)(i)and(1)(ii), the following investment function is incorporated.

where of investment(rate of capital accumulation), ρ=nominal rate of interest, πe=expected rate of price inflation, and ρ-πe=expected real rate of interest.Asada(2001)derived this type of investment function with debt effect from the expected profit maximization behavior of the firms by using the“principle of increasing risk”due to Kalecki(1937,1971), which means that firms'risk increases as the investment expenditure increases.

In the simplified model in this section, the fixed price is assumed so that we have πe=0.Furthermore, in this section, we assume that the central bank always acts to keep the nominal rate of interestρto be constant, which means the lack of the active monetary policy.Therefore, πe and ρ do not enter into the system of equations(1)as variables.

In this model, the change of the private debt-capital ratiobecomes an increasing function of the income-capital ratio yf12)>0, because the increase of y induces the increase of the debt financing through the rise of the investment expenditure g.On the other hand, the change of the income-capital ratiobecomes a decreasing function of the debtcapital ratio df12)<0, because the increase of d induces the decrease of the investment expenditure, which contributes to the reduction of the excess demand in the goods market through the decrease of the effective demand.These characteristics imply that we can consider the system of equations(1)as a“Lotka-Volterra”like system, in which the variable d plays the role of the predator and the variable y plays the role of the prey.

Asada(2001)proved that the cyclical fluctuations occur and in particular, the closed orbits around the equilibrium point exist at some range of the parameter valueαby means of the Hopf bifurcation theorem.See Gandolfo(2009)chap.24 as for the exposition of the Hopf bifurcation theorem.Figure 1 illustrates a typical closed orbit of income and private debt that is produced in this system and the time trajectories of two variables corresponding to four phases of the closed orbit.The cyclical fluctuations of income and private debt in Figure 1 can be considered as a typical example of the“Minsky cycle”.

Figure 1 Emergence of Minsky cycle(Source:Asada,2001, p.81)

Figure 2 illustrates two types of bifurcation that can emerge in this system.Here we omit the detailed interpretation how to derive Figures 1 and 2 because of the lack of the space.Figure 2(a)is an example of the“subcritical”Hopf bifurcation. In this case, the closed orbit exists in the range of the parameter valueαat which the equilibrium point is locally stable and the closed orbit itself is unstable.Figure 2(b)is an example of the“supercritical”Hopf bifurcation.In this case, the closed orbit exists in the range of the parameter valueα at which the equilibrium point is locally unstable and the closed orbit itself is stable.

Figure 2 Subcritical Hopf bifurcation(a)and supercritical Hopf bifurcation(b)(Source:Asada,2001, p.83)

Both of the above mentioned two types of Hopf bifurcation are economically meaningful.The“subcritical”case corresponds to the“corridor stability”in the sense of Leijonfufvud(1973), which means that this system is immune from small shocks that can contain the initial position inside the“corridor”(stable region), but it is vulnerable to large shocks that leave the initial condition outside the“corridor”.See also Asada(2010)as for the interpretation of the“corridor stability”.In the“supercritical”case, the time trajectory of the variables that starts from the initial condition other than equilib-rium point converges to the limit cycle, and the cyclical fluctuations persist indefinitely.