Estimation and Simulation of a Financialized Growth Regime with a Stock-Flow Consistent Model Luis REYES, Mickaël CLÉVENOT and Jacques MAZIER Introduction The French economy has gone through several liberal reforms during the eighties and nineties, particularly on the labor and the financial markets to face with the Fordist crisis. These reforms have set up a more financialized growth regime (Aglietta, 1998; Boyer, 2000). The so-called financialized growth regime is characterized by the research of a high return on own funds even at the expense of long term growth. These transformations of global behavior of French capitalism generate several problems; more macroeconomic instability, more unemployment, a weak growth despite the restoration of profits to a high level (the investment-profit puzzle). In this paper we try to describe the mechanisms which produced the macroeconomic instability of these growth regimes. This instability seems to be caused by wealth and leverage effects. This is the main reason why we are particularly interested in the financing mode and the financial structure of non financial companies (NFCs henceforth). The structure of NFCs is presented in the first part of the paper by means of several financial ratios which describe the increasingly important role played by the accumulation of financial assets at the expense of fixed capital accumulation. To show this aspect of financialization we present two behavioral equations of NFCs, which describe two ways of financing: an owns-funds norm and an indebtedness norm. These are estimated using modern statistical techniques for French NFCs using data coming from INSEE on a quarterly frequency from 1980 to 2009. These equations are then introduced in two different stock-flow models to carry out our simulations. These models, inspired by the methodology of Godley and Lavoie (2001), allow representing an overall financialized growth because the wealth and leverage effects are integrated in coherent social accounting matrices and equities prices are endogenized. In this 1 framework we know where all incomes (flows) come from and where they go to, in consumption or savings (stock). This aspect will be detailed in the second section. The results of simulations in the two configurations describe important financial cycles due to revaluation effects. In this way it is easy to create explosive cycles with small modifications of parameters in this kind of modeling but the main objective was to conserve their global stability to achieve demand or supply shocks, on the liquidity preference, equity issuing and to compare the behavior of the two regimes. The paper is organized as follows. We present the stylized facts in the first section; the framework of the stock-flow modeling exercise in the second section; some simulations of the two regimes in the third section and the econometric results for the specifications of the model will be described in the last section. I. The stylized facts of the financialization of the French economy Throughout the last thirty years the French economy, as many others, has undergone farreaching structural changes. These transformations are connected to the Fordist crisis. During the Fordist era (1950-1974), the growth regime was wage-led. The increases of wages induced a global dynamic demand combined with high productivity gains and global stabilization of income distribution. However, through time, the increasing protection of wage-earners led to a disconnection between wage and productivity gains. Moreover, globalization implied an increase of international competition. These two elements induced a reduction of profits and a drastic reduction of investment and growth. The growth regime turned les self-centered, competition coming from abroad became problematic, wages and salaries (up to then the main source of global demand) turned progressively more important as determinants of production costs. The less self-centered growth regime nowadays is, however, led by profits. To face this relative profit loss, and in order to regain control of inflation, some liberal reforms were set out. The liberalization and deregulation of many markets, in particular the labor and financial markets, have deeply transformed the mode of regulation and the accumulation pattern. The regulation mode dominated by wage-earners is now dominated by finance capital. 2 The profits of NFCs have come back to a high level without a return of a restoration of capital accumulation. The above-mentioned redistributive reforms finally led to a more financialized growth regime characterized by gloomy growth with more macroeconomic instability. The required level of financial profitability actually eliminated many productive investment projects and induced a rise of financial leverage to improve the return of own funds accompanied by a huge increase in risk. On the other hand, the trade-off between financial and physical return has led to more financial investment and less physical investment. Finally, the financial liberalization, justified by the increase of savings oriented towards physical investment, has not, to date, been materialized. On the contrary, liquidity in financial markets increases the cost of capital. Under a nominal point of view we can observe an increase of investment that is due to a huge increase of assets’ price without a direct linkage to real accumulation. We try to show the fundamental instability of the financialization process. This instability is an important characteristic of a financialized growth regime. It is due to the reduction of the rate of return in contrast to the required financial returns. As mentioned above, to show this aspect of financialization we present two behavioral equations of non financial companies which describe two ways of financing: an own funds norm and an indebtedness norm. These equations are then introduced in two different stock-flow models to build simulations. These simulations lead quickly to financial cycles for which it is easy to create explosive cycles with small modifications of the parameters. The paper is organized as follows. We present stylized facts in a first section, the framework of stock-flow modeling in a second section, econometric results for the two specifications of the model and some simulations of the two regimes will be described in the last section. Since the years following WWII two growth regimes distinguish themselves sharply in the French economy: the fordist regime and the financialized regime (Graph I.1). Under the Fordist regime the margin rate evolved around 29% on average, whereas the investment rate fluctuated around 23%. Thus, the rate of non-invested (financial) savings hardly exceeded 3 6%. From a high investment rate, along with strong growth ensuring the economy working at close to full employment, we are now facing a regime of weaker accumulation. Graph I.1 The succession of two growth regimes Since the second half of the eighties, France entered a new growth regime. This new accumulation pattern is characterized by low growth despite the restoration of margin and profit rates. This new configuration translates into an inversion of the Balance of power between labor and capital. During the last period of Fordism, wage-earners had more power in social negotiations, whereas financiers were relatively weak. With the two wars and inflation, the amount of financial holdings had been severely curtailed. The third group of members of this relation (managers) was essentially interested in increasing the size of their companies. The phenomenon of techno-structure described by Galbraith (1967) in the New Industrial State, allows us to explain a relative convergence of interest between workers and managers, the so-called rapport salarial fordien. But the excess of wage-earners’ demands for wage increases at the end of the seventies led to a classic profit crisis. The profit squeeze started at the beginning of the seventies due to the reduction of productivity gains of the Fordist organization of work. 4 A major element of the reconstruction of power of the financial sector was the enormous rise of interest rates at the end of the seventies driven by P. Volker in order to ‘fight’ inflation. This monetary policy led to a huge increase of the financial burden through the payment of interests (Graph I.2). This discouraged NFCs from incurring into debt, which generated a reduction of growth and accumulation and finally high inflation rates. But the nominal interest rate remained at a relatively high level that created a substantial rise of the real interest rate (Graph I.3). The situation of high real interest continued throughout the eighties and till the beginning of nineties. Graph I.2 Transformation of the financial burden In the recent period, the margin rate has increased to 31%, which contrasts with the 29% during the Fordist era. At the same time, the investment rate declined from 23 to 18% in the financialized regime. The share of profits not invested on physical capital has thus considerably increased. This rate has gone from 6 to 12% of gross operating surplus. The orientation of savings did not contribute to reinforce productive investment by reducing the cost of capital 5 but, on the contrary, it contributed to financial asset price inflation. This reinforced the cost of capital through higher demands coming from shareholders. This financial inflation translated into an important debt cycle for NFCs during the bubble of the New Economy. Then, it was households who, through increases in housing prices, will be the victim of a financial illusion. This mechanism will also be responsible of the housing crisis in Ireland, Spain and Greece. Graph I.3 The big rise of the real interest rate at the beginning of the eighties Henceforth, the financial sector as a crucial economic actor becomes again dominant as he had been able to be during the second half of the nineteenth century. The new balance of powers has destabilized the implicit agreement between managers and wage-earners. Finance imposes its main objective, the return on own funds which conduced to a reduction of physical accumulation A new distribution and a new orientation of income settled down. The requirements of physical returns on investments were sharply increased by decreasing the number of interesting projects (Stockhammer, 2004). As a result, capital accumulation has been reduced, growth has been lower and mass unemployment settled down in a long-lasting way in spite of some cyclical accelerations in the end of the eighties and 2000s. 6 These valuation effects appear in the structure of the flow of funds accounts of non financial companies or through Tobin’s Q (Graph I.6). The share of financial assets on the balance sheet has increased considerably. The debt ratio on own funds decreased in trend because of the rise of the financial value of companies. This ratio came from 80% at the beginning of the eighties to less than 40% in 2007. Graph I.4 Financial illusion There has been, indeed, a wealth effect. But these evolutions in the financial structure of NFC turn out relatively artificial because it does not represent exactly an increase of the capacities of financing of companies or their financial health. With the bubble burst in 2000 and, more importantly, in 2007 we can observe a destruction of value which leads to a new increase of this ratio. If we bring back the debts to the stock of productive capital, we can observe a degradation of financial situation of companies. Moreover, if we look at the debt-profit ratio, its level at the beginning of 1980 is the same as in 2008. 7 The financial situation of NFCs has gone back to its weak position of the first half of the eighties. The debt-productive capital ratio deteriorates, showing that the financial situation of firms has been hardly hit. But this deterioration is masked by a financial illusion induced by the increase in prices of financial assets. Financial illusion means that the financial structures describing a more important part of stockholders' equity at the level of the accounts of holdings, often hide a degradation of the operational accounts with a sensitive rise on behalf of the debts and of the debt servicing in the profits of companies (Graph I.4). Since the crisis of Net Economy we attend a sensitive degradation of the capacity of refinancing of the not financial companies. On the other hand, if this increase in financial value of NFCs had allowed increasing the financing of productive investment and had so been able to boost economic growth, the global situation would be less gloomy. But the financing by actions hardly represented 5% of the productive investment for the last 30 years (Graph I.5). The self-financing remains the privileged means of companies to finance their activity with a level of near 100% at the end of the 1990. The debts follow cycle with high variation. The net financing by financial markets represents less than 20% of investment. The instability of the financialized regime appears through the cyclic mattering movements of Q of Tobin and the volatility of the profitability of the equities (see below). This global instability is due to a wealth effect and financial leverage. These aspects are very well described by the stock-flow modeling presented which allows redrawing all the effects of stock (wealth/debts) on flows, financial burden, debt servicing and payments of dividends on all the economic relations. 8 Graph I.5 The weakness of equities to the financing of productive investment Graph I.6 Financial cylcle through the Tobin Q 9 To summarize and conclude this section, the settlement of a financialized growth regime in the French economy could be characterized by a few stylized facts: the investment-profit puzzle, the weakness and instability of growth, a weak commodities’ inflation combined with high unemployment and a high financial assets’ inflation. These stylized facts are consistent with the reversal of the balance of power between capital and labor which followed immediately after WWII. At the end of Fordist regime, workers’ bargaining power was rather strong, whereas in the financialized regime it is the shareholder’s value which has become too strong. In this era, the global distribution and using of income leads to a weak physical accumulation regime which implies problems of realization of value due to weak global demand. Demand has been sustained by debts which produced a high instability in Anglo-Saxon countries and the south of the European Monetary Union. In the French economy the relatively low uses in debt leads to a lower growth but with a less marked financial or estate crisis than elsewhere. II. Growth regime Fordism Financialized Growth Strong Weak Profit Normal High Unemployment Weak High Inflation High Weak Assets inflation Weak High A simplified Stock-Flow Consistent Framework II.1 Literature review To our knowledge, one of the first serious attempts to empirically deal with financial phenomena on a macroeconomic perspective, combining stocks and flows rather than dealing with one at the time, was that of Brainard and Tobin (1968), then extended by Tobin (1969) and others from the Yale school. This approach did not make its way to mainstream econom10 ics because it lacked micro foundations which explained the mechanism by which agents allocated their financial resources. When asked about the abandonment of SFC models, Tobin’s reply was “Well, people would rather do the other thing [computable, numerical or applied general equilibrium models] because it’s easier” (Dimand, 2003, p. 19). Some years later, a group of researchers from the Post Keynesian school took over SFC models and, thanks to the ease of access to large-scale reliable computational techniques rapidly evolving, gave them further solidity and more realism. Instead of a general equilibrium taste, these authors gave it a Keynesian/Minskyan flavor that aimed at explaining endogenously created disequilibrium without optimizing behavior from economic agents which, under this approach, is rather redundant and difficult to deal with in a realistic way. Lavoie and Godley (2001) and Godley and Lavoie (2007) account for the most influential works on this type of analyses. SFC models consist of systems of simultaneous equations which combine stocks (of debt, capital or deposits) with flows (of production, income or liquid assets) using experimental (simulated) data in a realistic accounting framework. Considering the fact that all form of wealth (including capital gains and valuation effect) in an economy comes from somewhere and goes somewhere, these models have an advantage above other techniques: they are capable of describing the mechanism underlying a shock, either coming from the financial or the real side of the economy, and its effect on macroeconomic aggregates. They are especially well suited to study a finance-led growth regime. Just as in Brainard and Tobin (1968, p. 363) a priori1 “The numerical values [of the parameters] embody some preconceptions of the authors”. However, it must be noted that these parameters (at least most of them) are based on economic theory and, although still subject to refutation, are accepted by a considerable number of important academicians. In order to distinguish a model explaining one economy from another, initial values are set for the variables used in the model based on real data from the economy to be studied. In the present case, these starting values are approximated for France in fiscal year 2009. Moreover, for simplicity we study France as it were a closed economy under certain simplify1 The aim is, in further research, to integrate official data sources into the model. A first attempt is here made and we present both econometric results and a simulation which, though we do not use the exact numerical values, we did find the “right signs” for accumulation (physical and financial), the debtcapital ratio and an own-funds norm. 11 ing (still realistic) assumptions which will be made clear when needed. Thus an interesting extension will be to make this model in an open economy context. Another important extension is to use econometric techniques to find the corresponding parameters of the equations, as it is done below2. In Minsky’s (1986) model, the surge in investment in bull phase of the business cycle is allowed by an increase in external financing (debt only in the present model) which explains the endogenous fragility of firms, i.e. the increase in default risk. In the ascending phase of a cycle, the reduction of investors’ liquidity preference on financial markets, that is to say the decrease of the risk perceived by the investors, allows the increase of the debt share in firms’ balance sheets. Firms thus take advantage of this situation to increase their financial leverage. But this process ends because of an endogenous reversal of the liquidity preference which corresponds to a reversal of collective opinion on financial markets. As a consequence, credit risk is revised upward, which generates the fall in investment. When investors on financial markets begin to have doubts about the value of collateral (the sum of retained earnings here) liquidity preference starts rising and this generates a fall of the prices on financial markets. These doubts generate a revaluation of credit risk. Investors run towards liquidity, which thus leads the firms to run strong insolvency risks since the refinancing of debt becomes extremely difficult. The Stock Flow Consistent (SFC) approach is well suited to analyze these questions. Thanks to a complete description of the balance sheets of each agent and of the associated flows of funds, the main components of Post-Keynesian macroeconomic models can be incorporated in a consistent way: relations between capital accumulation and income distribution, wealth effects (especially for rentiers), valorization effects (due to capital gains or losses), and a debt-led regime with Minskyan perspective. In a convergent way, Lavoie and Godley (2001), Godley and Lavoie (2007), Taylor (2004, 2010), Dos Santos and Zezza (2008) have proposed SFC models including most of these factors. Although close, these models differ in some points. Godley and Lavoie use computer simulations to study the nature of the growth regime while Taylor, Dos Santos and 2 This is preliminarily done for debt and equity, and for real and financial accumulation following the estimations of Clévenot, Guy and Mazier (2010), and for investment in Guy, Clévenot and Mazier (2010), but not for other equations. 12 Zezza study analytically the dynamics of their models. Beyond this methodological divergence, the models differ in the way they deal with debt and equity issuing. These are actually two alternative closures of the model to represent how firms finance capital accumulation, which we consider in our simulations. Godley and Lavoie, Dos Santos and Zezza, as well as Taylor in some of his models, retain an equation describing the issued equities. Consequently, credit demand by firms is simply determined as a residual of the firms’ financing account. In Taylor (2010) asset prices display positive feedback but must eventually be reversed by other forces. The growth rate of asset prices depends positively on the return to equity and the valuation (value of equitiescapital stock) ratio, and negatively on the dividends-capital stock ratio. The growth rate of the number of equities depends positively on the accumulation rate and the share of newly issued equities on the capital stock, and negatively on the valuation ratio. Growth of the capital stock can stabilize the valuation ratio which affects negatively equities, but positively their price. Alternatively, Taylor (2004), in two other versions of his models, retains an explicit firms’ credit demand equation with no issued equities or with equities determined as a residual of firms’ budget constraint. These questions are not discussed in details in the SFC literature and may not be central for models’ properties. But this arbitrage between debt, equity and retained profit is important in the growth regime which prevailed since the 1990s. In this perspective, a simplified SFC framework will be outlined with two model’s versions corresponding to the main closures previously discussed, one with an indebtedness norm or loan demand, the other with an own funds norm or issued equities. II.2 The model There are five sectors in the economy: households, non financial firms, the government, private banks and a Central Bank. The price level is assumed to be constant across all periods. The price of equities plays a market-clearing role. Table 1 shows the matrix of stocks. The first column describes the stocks of wealth held by households (Vh), which is made up of cash (Hh), bank deposits (BD), bonds (pb*B, where pb is their price) and equities (pe*Eh, with pe the price of equities). In the same vein, firms contract debts (L), hold equities (pe*Ee) and issue equities (pe*E) in order to finance capital accumulation (K), and they hold an outstanding amount of wealth (Ve). The government issues the bonds households retain and Treasury 13 bills held by banks (BT). Total government debt (Vg) is the sum of both with a minus sign. Private banks hold a total amount of wealth (Vb) which comes from holding reserves (Hb), receiving deposits from households, making loans to firms, lending to the government and getting refinanced by the Central Bank, which in turn issues all the Central Bank money (H) and holds no wealth. Table II.1 Matrix of Stocks Households Firms Government Banks Central Bank Hh Hb -H BD -BD K -L pb*B pe*Eh L -pb*B pe*Ee -pe*E -BT BT -RF -Vh -Ve -Vg RF -Vb The first equation of the model is the national income identity and, as we assume a closed economy, the equation says that national income is equal to the sum of consumption, investment and government spending3: (1) Y=C+I+G Households’ behavior Equations (2)-(11) describe households’ decisions. Disposable income (YDh) is the sum of wages (W), interests on bank deposits and on bonds one period before, and dividends (DIVh) net of taxes (T). The Haig-Simons definition of income is the sum of disposable income and 3 Government spending is assumed to grow at a constant rate. 14 capital gains of households (CGh). Taxes are a proportion (θ) of gross disposable income. The standard Keynesian consumption function depends on the Haig-Simons definition of income, where a0 is autonomous consumption and a1 is the marginal propensity to consume, and on a (lagged) ‘wealth effect’ described by a2. (2) YDh = W + id*BD-1 + B-1 + DIVh – T (3) YHSh = YDh + CGh (4) T = θ*(W + id*BD-1 + B-1 + DIVh) (5) C = a0 + a1*YHSh + a2*Vh-1 Following Godley and Lavoie’s (2007) approach, bonds, as a proportion of households’ wealth, is a linear function of the interest rate on bills (rb), the interest rate on deposits (id) and the rate of return on issued equities (re), with the last two affecting it negatively. The proportion of the value of equities held by households (pe*Eh) out of their total wealth is negatively influenced by the interest rates and has positive own feedback through its rate of return. The cash held by households are a proportion (λ0) of consumption. The change (Δ) in bank deposits is calculated as a residual of other forms of incoming wealth. Capital gains of households are defined by the change in the prices of the bonds and equities they hold multiplied by their corresponding amounts lagged one period. Total households’ wealth was defined above. (6) pb*B = Vh*(v0 + v1*rb – v2*id – v3*re) (7) pe*Eh = Vh*(w0 – w1*rb – w2*id + w3*re) (8) Hh = λ0*C (9) ΔBD = YDh – C – pb*ΔB – pe*ΔEh – ΔHh (10) CGh = Δpb*B-1 + Δpe*Eh-1 (11) Vh = BD + pb*B + pe*Eh + Hh Firms’ behavior Firms’ decisions are described in equations (12)-(27). Following a Kaleckian framework, the investment function (eqs. 12-14) is assumed to depend positively on the lagged profit rate (UP/K-1) and the growth rate of the economy (ΔY/Y-1) with k2 being the accelerator effect. It 15 depends negatively on the debt ratio (L-1/K-1), according to an increasing risk effect, and on the interest rate on loans. Last, the financial rate of return on equities held (ree) has also a negative impact, reflecting an arbitrage between real and financial accumulation. δ is the depreciation rate of capital. (12) g = k0 + k1*(UP-1/K-2) + k2*(ΔY/Y-1) – k3*(L-1/K-1) – k4*rl –k5*ree (13) I = g*K-1 (14) ΔK = I – δK-1 The financial accumulation is described through the share of the value of equities held by firms out of their total capital (real and financial). It is a linear function of the rate of return on equities held (ree) and the profit rate. An alternative specification will focus directly on the rate of financial accumulation (peΔEe/ (peEe)-1) explained by the same variables (financial rate of return ree and profit rate), but also by the debt ratio, with a positive influence as a leverage effect favors financial accumulation, in contrast with the negative impact of the increasing risk on real investment. (15) pe*Ee = (K + pe*Ee)*(f0 + f1*ree + f2*(UP/K-1)) (15bis) peΔEe/ (peEe)-1 = f0 + f1*ree + f2*(UP/K-1) +f3* (L-1/K-1) Two alternative closures Equations (16) and (16bis) are, respectively, the debt ratio and the own funds norm equations, which are used alternatively in each model. Thus, Model 1 uses (16), and the amount of equities issued (E) is deducted from (17), solving for ΔE. Similarly, Model 2 uses (16bis), and debt is deducted from (17), solving for ΔL. We proceed in this fashion to analyze Minskytype cycles when firms finance investment by external funds (debt) and by internal funds (undistributed profit or issuing equities). The left-hand side of (17) is the spending decision of firms (between investing and holding equities), whereas the right-hand side represents their income (from profits, from issuing equities and from contracting loans). The debt ratio, interpreted as an indebtedness norm (eq. 16), depends positively of the rate of profit, as higher profitability makes easier to get loan from banks, of the rate of return on equities, as a higher cost of issued equities makes credit more attractive, and last, as usual, of the rate of interest. 16 Conversely, the own funds ratio, measured in percent of the total real and financial assets (eq. 16bis), depends positively of the interest rate, as a higher credit cost makes equities issuing more interesting, of the debt ratio, as an increase of the indebtedness pushes firms to use more internal funds, and last negatively of the rate of return of equities, as a higher cost of issued equities makes issuing them less interesting. (16) L/K = g0 + g1*(UP/K-1) + g2*re – g3*rl (Model 1) (16bis) pe*E = (z0 + z1*rl + z2*(L-1/K-1) – z3*re)*(K + pe*Ee) (Model 2) (17) I + pe*ΔEe = UP + pe*ΔE + ΔL Undistributed profits (UP) are deducted from total income minus costs, which in this case account for wages, interests and households’ dividends. Wages are a constant (r0) share of income. The rate of return of equities issued is equal to the share of the capital gains (or losses) augmented of the distributed dividends, in percent of the total equities issued, which is equivalent to the growth rate of the price of equities plus the share of distributed dividends (DIV) out of total equities issued. Dividends, in turn, are calculated (as in Lavoie and Godley, 2001) as a proportion (1 – sf) of profits realized the previous period. Dividends paid to firms (DIVe) are here defined as the share of equities held by firms out of total equities issued in the previous period (Ee-1/E-1). Dividends paid to households, as well as their equities, are calculated as a residual. Firms’ capital gains (CGe) come from changes in the price of equities multiplied by the amount held by these enterprises. The outstanding amount of wealth held by firms was defined through the matrix of stocks. (18) UP = Y – W – rl*L-1 – DIVh (19) W = r0*Y (20) re = (Δpe/pe-1) + DIV/(pe-1*E-1) (21) DIV = (1 – sf)*(Y-1 – W-1 – rl-1*L-2) (22) DIVe = DIV*(Ee-1/E-1) (23) DIVh = DIV – DIVe (24) Eh = E – Ee 17 (25) CGe = Δpe*Ee-1 (26) Ve = K + pe*Ee – L – pe*E Government Equation (27) describes the government’s issue of Treasury bills (ΔBT), which is calculated as a residual of its expenditures–on current spending, interests on treasury bills and bonds– and its revenues –from taxes on personal income, taxes on banks (TB) and taxes on the Central Bank (TCB) and from new issued bonds (pb*ΔB). The price of bonds is assumed to vary inversely with respect to the interest rate paid, which is assumed to be equal to interest rate on bills (short-run). Total wealth held by the government is equal to its debt. (27) ΔBT = G + r*BT-1 + B-1 – T – TB – TCB – pb*ΔB (28) pb = 1/rb (29) Vg = – D = – BT – pb*B Banking sector Private banks make profits (BP) and pay taxes (TB) out of their income, which is made up of interests on loans to non financial firms and to the government minus interests paid on deposits and refinancing from the Central Bank. θ is the tax rate they pay. Their refinancing (RF) comes from their expenditures, mandatory reserves (Hb) at the Central Bank, loans and Treasury bills, minus their retained profits and deposits they receive. This refinancing is granted without restriction by the Central Bank. Mandatory reserves are a proportion (λ) of bank deposits. The change in wealth held by them (ΔVb) is their profits. (30) BP = (1 – θb)*(rl*L-1 + r*BT-1 – id*BD-1 – ib*RF-1) (31) TB = θb*(rl*L-1 + r*BT-1 – id*BD-1 – ib*RF-1) (32) ΔRF = ΔHb + ΔL + ΔBT – BP – ΔBD (33) Hb = λ*BD (34) ΔVb = BP 18 The Central Bank receives interests from private banks for previous refinancing and transfers them as tax to the government (TCB). Consequently, the Central Bank makes no profit and its net wealth remains constant, equal to zero. Total high power money (H) is the sum of cash held by households and reserves made by commercial banks. The interest rate on loans (rl) is assumed higher than the short term interest rate controlled by the Central Bank (ib) and supposed exogenous, where m1b ( m2b) is the spread. Inversely, interest rate of deposits (id) is supposed smaller, which is at the origin of banks’ profit. Interest rate on Treasury bills ( r) is assumed to be equal to interest rate on loans (rl), which is in turn equal to the yield on longterm bonds (rb). (35) TCB = ib*RF-1 (36) H = Hh + Hb (37) rl = ib + m1b (38) id = ib – m2b (39) r = rl (40) rb = r In order to make sure that in the model all flows come from somewhere and go somewhere, we make sure that in both models H = RF (the Central Bank’s equilibrium, the hidden equation which is not written). The final condition for the model to be coherent, is that the capital stock be equal to the sum of all wealth held by all the economic agents in the model; Vh + Ve + Vg + Vb = K. The working out of the model Table II.2 summarizes in a simplified way the main determinants of fixed and financial accumulation on one hand, of equity issuing and indebtedness on the other, as they result from the outlined SFC model. These relations characterize some of the main features of the finance-led 19 growth regime regarding firms. They allow us to describe financial cycles following a Minskyan perspective, as it is illustrated in Figure II.14. Table II.2 Main financial determinants of firms’ behavior Explained variables Explaining variables Fixed capital Financial accumulation accumulation Equity issuing Rate of profit + + Real interest rate - - + Debt ratio - + + - + - Financial rate of return Debt ratio + - + Note: Signs of partial derivative of explained variables regarding each explaining variable according to each equation Starting with a rising financial rate of return, financial accumulation is stimulated while equities issued are reduced. This leads to increasing equities’ price as a clearing market result, which accentuates the initial rise of the financial rate of return. On the other hand, higher financial profitability induces firms to borrow more and increase their indebtedness, which, through the leverage effect, sustains financial accumulation. In this environment fixed capital accumulation is slowed down through negative impacts of the rise of both, the financial rate of return and the debt ratio reflecting an increasing risk. The contrast between booming financial accumulation and limited recovery of fixed capital accumulation has been a figure of the nineties and 2000s in France. In this ascending phase of the financial cycle, there is no stabilizing mechanism, except the positive effect of rising indebtedness which leads firms to issue 4 The two closures of the model are presented simultaneously in the Graph 7 for sake of simplification, although equities issuing or debt is alternatively determined as a residual through an accounting equation. 20 more equities, contributing to limit the increase of equities’ price. In that sense the model does not describe how the process can end in an endogenous reversal, which can reflect the instability of this growth regime. Simulations in the next section will help to clarify this question. Figure II.1 The interaction between the main firms’ parameters in the framework of a financial cycle + Equity Issuing Debt ratio - peΔE - + - Financial profitability Equities price - + re pe Productive investment Financial accumulation pkΔK + peΔEe + + Three points can be added. First, a restrictive monetary policy can contribute to stabilize the system. Increasing interest rates slow down financial accumulation while it favors equities issuing, which helps to depress equities’ price and financial profitability. On the other hand, indebtedness is reduced which also limits financial accumulation. But the overall effect on fixed investment and growth is, most of the time, negative due to the rising cost of credit. Second, the economic environment and the demand side can be taken into account using the SFC model. Rising equities’ price induces capital gains and increase households’ wealth, which sustain their consumption and, indirectly, the demand and fixed investment. 21 Higher rate of profit stimulates both fixed investment and financial accumulation and authorizes more recourse to indebtedness, which indirectly favors equities issuing and contributes to stabilize the system. Third, this model only focuses on relations between firms and finance, which is an important component of the finance led growth regime. But it gives a simplified representation of households, as it ignores their indebtedness and their capital accumulation in housing, which has plaid a crucial role at the origin of the current financial crisis. Households’ portfolio behavior would also have to be adapted with two kinds of households, according to the level of theirs wealth and incomes. Last, banks’ behavior is also highly simplified and doesn’t reflect their active role, both in financial accumulation and in financialization. However, some new and improved econometric results will be given regarding the relations between firms’ capital accumulation and finance. A first set of preliminary simulations of the SFC model will be proposed in the last section and will give a better understanding of the working of the model. III. Simulations: preliminary results Two models will be examined, Model 1 with an indebtedness norm and Model 2 with an ownfunds norm5. In order to study the mechanisms of these two models, shocks on the demand and supply sides and on the financial side will be carried out. Before proceeding with the description of these shocks, it seems useful to show the baseline GDP growth rate reference for each model. As shown in Graph III.1, Model 1 with the indebtedness norm exhibits 5-period cycles (from peak to peak) which vanish over time. On the other hand, Model 2 with an ownfunds norm also shows cycles, but over a much longer period (from peak to trough there is more than 50 periods), as shown in Graph II. 2. The nature of these contrasted cyclical behaviors will be better understood below thanks to the shocks and the adjustment mechanisms which will be analyzed in more details. 5 It must be noticed that the simulations presented here have been done with a calibration based on ecometric estimations slightyly different from those of the part 3 of the paper. An actualisation with the new estimations wwill be done in a new versión of the paper. 22 Graph III.1 Output growth path of Model 1, indebtedness norm. Graph II.2 Output growth path of Model 2, own funds norm We now carry out our simulation experiments, which consist in shocking five variables out of our system: the consumption function, the wage share, the investment function, the demand for equities from firms, and the demand for equities from households. Shocks 1 and 2 are on the demand side, Shock 3 on accumulation, and Shocks 4 and 5 on the financial side. The 23 effects of these shocks are analyzed graphically for Model 1 (indebtedness norm) and Model 2 (own funds norm) on the following variables: output (Y), personal consumption (CP), the price of equities (pe), the accumulation rate (I/K), the profit rate (UP/K), the share of equities held by firms out of their total assets (pe*Ee/(K+pe*Ee)), the debt ratio (L/K), the wealth effect (Vh/YDh), and the financial rate of return (re). We assume a once-and-for-all change in period 31, and each shock will be compared to the corresponding baseline solution (see Graphs II.1 and II.2 above). Although shocks run from t = 31 to 60, the reader must bear in mind that what we analyze here are once-and-for-all shocks on single variables, which in turn imply no other change in economic policy or other exogenous factors. The possibility of policy responses is also left for further research. III.1 Increase of households’ consumption Model 1 with indebtedness norm We begin by describing a shock on the demand side. We assume that autonomous consumption increases 2.5% out of total personal consumption. That is, a0 = 0.566 increases to 2.066, or Δa0 = 1.5. Graph III.3 illustrates what happens in Model 1 with the indebtedness norm. The top-left graph is total output given Shock 1 (Y_1) divided by total output in the corresponding baseline scenario (Y_0). The top-right graph is the same ratio for personal consumption, followed by the price of equities. The following graphs are differences of the corresponding after-shock variables and the baseline for the investment-capital ratio, the profit rate, equities held by firms, the leverage ratio, the households’ ‘wealth effect’ and the rate of return on issued equities, respectively. As can be seen from the Graphs, an increase in personal consumption has the expected positive effect on output which, although less than proportional, takes place immediately and extends to the longer-run, following a traditional Keynesian recovery. This brings about an increase in the price of equities, as firms issue less equities thanks to the economic recovery and the improvement of undistributed profits. Two periods after the rather strong positive effects of this shock take place, the first economic downturn occurs (top left graph), which is then followed by a milder three period trough. The price of equities reaches its peaks one period after output does. This downturn of the price of equities is the consequence of the fall in output and profit which induces new issue of equities facing the indebtedness norm. With the slowdown investment declines and firms reduce their equities issuing. This allows a new up24 turn of equities price. Consequently a financial cycle can be observed but business cycles become progressively milder. The other variables give more information. The rate of accumulation (I/K) decreases slightly in the first period after the increase in autonomous consumption, due to the improvement in the financial rate of return which has a negative effect. But it then increases significantly with the recovery. For up to four more periods until profits fall enough for firms to begin issuing equities, which again makes output fall. These differences then become less and less important. For the same reasons just described, the rate of return on equities held and the rate of financial accumulation evolve cyclically. With the indebtedness norm, fluctuations of the debt ratio remain limited. On the whole, financial cycles can be observed at the level of equities’ market, with acceleration and slowdown of equities’ prices growth, and at the level of the financial rate of return. This is mainly explained by the variation of equities issued facing the constraint of financing with the indebtedness norm and by the role plaid by equities’ prices to clear the market. Model 2 with own funds norm The same shock is carried out with Model 2 (own funds norm). Graph III.4 also shows at short term a positive effect of an increase in personal consumption on output, although of more limited amplitude than in Model 1. The price of equities, the profit rate, the equities held by firms and the rate of return on equities also increase. At short term the debt ratio slightly decreases, compared with the base line, contrary to what was observed with the Model 1. In Model 2 with an own-funds norm, loans to firms are determined as a residual. in the shortterm firms need less credit thanks to the improvement of the profit with the recovery and to the preservation of equities issued with the own funds norm. But at medium term, the evolution is quite different. There is a financial bubble with higher financial rate of return, increasing financial accumulation and a permanent decline of the real rate of accumulation. Firms’ indebtedness increases without limit which stimulates the financial accumulation and the growth of equities’ prices but reinforces the slowdown of investment and of the production. 25 Graph III.3 Shock increase of consumption for Model 1, indebtedness norm. Indeed, the two versions of the model show contrasted mechanisms. With Model 1 and the indebtedness norm, there are short-term financial cycles with equities issued determined as a residual and equities’ prices clearing the market. With Model 2 and the own funds norm, there is on the contrary a financial bubble with increasing financial accumulation and equities’ pric- 26 es. There is no stabilizing mechanism. Loans are determined as a residual and the debt ratio increases without limit. Graph III.4 Shock 1 increase of consumption for Model 2, own funds norm. 27 III.2 Increase of the wage share Shock 2 on the wage share is assumed, in which r0 increases from 0.67652 to 0.69 (Δr0 = 0.01348), or roughly 1.9%. Graphs III.5 and III.6 show the results for Models 1 and 2. Model 1 with indebtedness norm With the indebtedness norm the increase in the wage share implies lower output because investment is sensitive to the fall of the profit rate and consumption does not increase to offset this fall in output. The fall of profit and the constraint of the indebtedness norm push firms to issue more equities which are determined as s residual. This induces a decline of equities’ prices to clear the market and a drastic decline in financial profitability. Indeed consumption falls despite the increase in the wage share, because capital gains of households fall6. The slowdown of the activity and of the rate of accumulation reduces later on the issued equities, which contributes to stabilize the equities’ prices and the financial rate of return. Consequently, a financial cycle can be one again observed with a debt ratio fluctuating moderately with the constraint of the indebtedness norm. On the whole, as the model is calibrated, the economy appears profit-led with financial fluctuations. Model 2 with own funds norm On the contrary, Model 2 with the own-funds norm appears wage-led in the short- to medium term (Graph III.6). It shows that the switch from capitalists’ income to workers’ income implies a short- to medium-run increase in output more in line with the Post-Keynesian wage led tradition. To offset the declining rate of profit, firms now get more indebted, with an increasing debt ratio, as loans are determined as a residual in this Model 2. This contributes to limit the fall in investment. It also sustains a financial accumulation with increasing equities’ prices. In the longer-run the decrease in investment weighs on output growth, which, in the absence of any appropriate policy response, falls. On the whole, the opposition between the two models is confirmed, the Model 1 with indebtedness norm is more driven by financial cycles with equities’ prices clearing the market while in Model 2 with own funds financial accumulation with increasing equities prices are at work. 6 The wage share increase has a negative effect on consumption under this specification, but the reader must be aware that this is due to the important amount of equities in households’ wealth. This should be further improved. 28 Graph III.5 Shock increase of the wage share for Model 1, indebtedness norm. 29 Graph III.6 Shock increase of the wage share for Model 2, own funds norm. 30 III.3 Increase of the accumulation rate Graphs III.7 and III.8 show what happens to the economy under a shock which implies a 0.5% increase in the rate of capital accumulation (Δk0 = 0.005). Model 1 with indebtedness norm Starting with Model 1 (indebtedness norm), this demand shock implies a permanent increase in output driven by investment and a permanent decrease in the price of equities. The consequent decrease in financial profitability keeps investment from falling, which in turn makes the capital stock grow proportionally more than undistributed profits, thus gradually reducing the profit rate. This is explained by the fact that firms are constrained by their indebtedness norm and issue more equities which, following an insufficiently increasing demand for equities, makes their price decrease. In the medium-run, financial accumulation by firms is reduced due to the worsening of the rate of return on equities issued. Demand is sustained by consumption and investment at the expense of capitalists’ income coming from both the real and financial sides. In this shock the financial cyclical behavior remains with the clearing market role plaid by equities prices but is partly offset by the general growth trend. Model 2 with own funds norm In Model 2 (own funds norm) the shock on investment is longer-lasting in the economy (Graph III.8). The price of equities increases, due to the own funds norm which limits their supply and this in turn implies an increase in the financial rate of return which sustains financial accumulation and the financial bubble. Firms’ indebtedness grows so as to finance supplementary real and financial investment. This growth of the debt ratio is without limit as loans are determined as a residual and can be obtained without restriction. The increase in the price of equities brings about capital gains capable of holding demand at high levels in spite of a decreasing rate of accumulation in the long run. This fall is due to the sensitiveness of the investment function to the negative effect of financial profitability and of the debt ratio, as seen in equation 12. The profit rate remains higher than in the corresponding baseline solution but accumulation eventually falls in the medium-run, both as a consequence of the financial boom and of the increasing indebtedness. Growth at long term is sustained by households’ 31 consumption which benefits of wealth effects7. This shock illustrates a combination of a finance-led growth with increasing indebtedness. This shock on the accumulation rate gives another illustration of the opposition between the two models. In Model 1 with indebtedness norm growth is mainly driven by the investment with limited financial accumulation and declining financial rate of return. The financial cyclical behavior remains under constraint thanks to the general growth trend. In Model 2 with own funds norm growth is more finance led with a financial bubble and increasing indebtedness which limits investment in the long run but supports growth thanks to wealth effects. 7 Here again this wealth effect could be revised in other calibration reducing the amount of equities held by households. 32 Graph III.7 Shock increase of the accumulation rate for Model 1, indebtedness norm. 33 Graph III.8 Shock increase of the accumulation rate for Model 2, own funds norm. 34 III.4 Increase of firms’ financial rate of accumulation Graphs III.9 and III.10 are for Models 1 and 2. They assume a 5% increase of the financial rate of accumulation in equation (15) with Δf0 = 0.103. Model 1 with indebtedness norm In Model 1 with indebtedness norm the financial shock on firms’ demand for equities, implies a cyclical increase in output (of limited amplitude), thanks to a stock market boom seen through the increases of equities’ prices and of the financial rate of return. Capital gains stimulate households’ demand. However, it is followed by a downturn of financial profitability due to the new equities issued by firms, which are a consequence of their indebtedness constraint. Troughs are not as deep so as to erase initial gains and variations of profit rate remains above the variations of the rate of accumulation. A financial cycle is observed later on, as in the previous shocks, with equities’ prices clearing the market. Model 2 with own founds norm Model 2 with own funds norm presents a paradoxical result, also found for the next shock (see below), originating in the financial side. The paradox here is that both real investment and financial investment become less attractive (as seen in their corresponding profitability rates) but there is an increasing growth rate for the economy. There is, however, an initial decline in output due to the drastic decrease of the price of equities (linked to the increase in accumulation by firms and to the own funds norm). It is followed by a recovery of output growth due to the lowering of the financial rate of return, which in turn makes real investment more attractive in the medium- to long-run. Thus the initial decline in the profit rate is followed by a further recovery. Although accumulation increases significantly at short term, it tends to fall in the medium- to long-run because of the increasing debt ratio. 35 Graph III.9 Shock increase of financial rate of accumulation by firms for Model 1, indebtedness norm. 36 Graph III.10 Shock increase of financial rate of accumulation by firms for Model 2, own funds norm. 37 III.5 Increase of the demand for equities from households Graphs III.11 and III.12 describe for Models 1 and 2, respectively, a shock assuming an exogenous increase in the demand for equities from households (as a proportion of their wealth) of 1% (Δw0 = 0.004). Model 1 with indebtedness norm This financial shock on households’ demand of equities generates large financial cycles with a succession of financial crises. Initially equities’ prices are boosted by the stronger demand which increases the financial rate of return. Capital gains improve households’ income and demand while firms’ investment is reduced to the benefit of financial accumulation. However, a reversal appears a few periods later. Facing the indebtedness norm, equities’ issue increases, which depress the financial market and induce a decline of equities’ prices and of the financial rate of return. This in turn has a negative impact on households’ income and demand and, more broadly, on growth. Financial cycles follow as before, but with more intensity than in the previous shocks. In the longer-run, the increase in the price of equities is unsustainable and thus tends to fall despite the peaks which occur every 5 periods. This happens because non financial firms must issue more equities to finance investment, due to the indebtedness constraint they face. Broadly speaking, what we see is a succession of financial cycles with similar effects on the real side of the economy without any gain in terms of output growth in the medium- to long-run. Model 2 with own funds norm For Model 2 with the own funds norm (Graph III.12) the supplementary demand for equities from households gives results close to those observed in case of increasing financial accumulation by firms (see above). It induces a short-term decrease in output and a decrease of the price of equities. Consumption falls more drastically than output, deposits are reduced significantly, but dividends for households double every 6 periods, due to the drastic decline in firms’ demand for equities, which firms issue almost exclusively for households. Output closely follows the evolution of the share of equities held by firms on their own funds. The rate of accumulation grows mainly due to the decline of the financial rate of return which induces firms to invest more in real capital. Losses are absorbed mostly by capitalists who see 38 their profit rate reduced. This increase in disposable income, through households’ dividends (which grow drastically), appears hardly sustainable in reality. Graph III.11 Shock increase of demand for equities from households for Model 1, indebtedness norm. 39 Graph III.12 Shock 5 increase of demand for equities from households for Model 2, own funds norm. 40 III.6 Final remarks and conclusion of the section We have studied a “finance-led” growth regime, which consists of a financialized system where productive and financial investment affect (and are affected by) the real side of an economy. For such purpose, we have used a Stock-Flow Consistent model with five sectors: households, non-financial firms, government, private banks and the Central Bank. All stocks and all flows come from somewhere and go somewhere. Two alternative closures of the model have been proposed, one with an indebtedness norm where issued equities are determined as a residual, and another with an own funds norm where loans to firms are on their turn determined as a residual. Simulations with shocks on the demand side or on the financial side have helped to give a better understanding of the working out of the model. Indeed, the two versions of the model have shown contrasted mechanisms. In Model 1 with indebtedness norm, there are short term financial cycles with equities issued determined as a residual for the need of real and financial investment. Equities’ prices are clearing the market. Consequently, financial fluctuations with upturn and downturn more or less pronounced according to the cases are the normal mode of regulation of this financial regime. On the opposite, in Model 2 with the own funds norm, there is a financial bubble with increasing financial accumulation and rising equities’ prices or a permanent financial deflation according to the cases. There is no stabilizing mechanism. Loans are determined as a residual and the debt ratio increases or decreases without limit. This financial regime appears structurally instable. These results have appeared clearly, both in the shocks on households’ demand and on the wage share. The shock on the accumulation rate has given another illustration of the opposition between the two models. In Model 1 with indebtedness norm growth is mainly driven by the investment with limited financial accumulation and declining financial rate of return. The financial cyclical behavior remains under constraint thanks to the general growth trend. In Model 2 with own funds norm growth is more finance led with a financial bubble and increasing indebtedness which limits investment in the long run but supports growth thanks to wealth effects. Shocks on the financial sector, on firms’ financial accumulation or households’ equities demand, have confirmed the previous observations in the case of Model 1 with indebted- 41 ness norm. Financial cycles with a succession of financial crises are observed in both cases. On the contrary Model 2 with own founds norm has appeared more paradoxical. These results must be regarded as preliminary. It would be useful to check the robustness of these conclusions according to the specifications used to characterize the two types of indebtedness norm or own funds norm. The importance of the wealth effect in households’ behavior is another factor to examine. Last, the confrontation with the empirical observation of the French case during the period 1990-2012 would also have to be made more in detail. IV. Econometric estimations for France All estimated equations run from the first quarter of 1980 to the fourth quarter of 2009. Annual capital (K), free of depreciation, for French nonfinancial firms comes from INSEE’s Comptes de Patrimoine and was brought to quarters using the Denton (1971) method (see Appendix for data sources). Unless otherwise specified, we use productive capital 8 for the construction of the financial ratios including this variable. The series used to approximate the annual series to quarters was Gross Fixed Capital Formation in volume. All series are smoothed using simple or double moving averages in order to avoid seasonal dependence and volatility, thus reducing the chances of having autocorrelation and heteroskedasticity in the models. All data are for nonfinancial firms, except utilization (which is proxied by the rate of capacity utilization from the manufacturing sector) and the real long-run interest rate. The profit rate (EBE/pkK-1)9 is here defined as gross operating profit divided by the stock of capital from the previous period at the current capital price. The indebtedness ratio (L/pkK) is defined as debt divided by the stock of capital, where debt is defined as the total liabilities less equities issued (which evolves essentially the same as credit and credit plus other account payable). The financial rate of return of equities held (ree) is, as in Clévenot, Guy and Mazier (2011), equal to the growth rate of the price of equities plus the ratio of received dividends and the stock of equities held the previous period. 8 We deducted the value of land and the corresponding rent from capital, both as a stock and as a flow, see Appendix for more details. 9 We omit time subscripts for simplicity, as in Godley and Lavoie (2001). 42 The series used in each model are integrated of order 1 (ADF tests shown in the Appendix), mainly due to the presence of a changing variance and not only, as it is frequently and mistakenly believed, due to the presence of a deterministic trend. Therefore, our series are difference stationary, as opposed to trend stationary. Misspecification tests (both at individual equation- and system-wide levels), which play a crucial role to test the validity of statistical inference, are presented in the Appendix. The lag order of the models was decided upon on the basis of their post-estimation statistical significance. IV.1 Accumulation The first estimated VAR(6) (with a restricted trend, not shown = 0.0014) says that, in the long-run, capital accumulation depends positively on the profit rate and utilization, and negatively on indebtedness, the interest rate and the financial rate of return. Given the number of lags, the sample size for this model goes from 1981(3) – 2009(4), which accounts for 114 observations. A total of 240 parameters (from 6 equations with 6 lags plus a constant, a trend and two dummies each) where estimated and summarized in the long-run relationship: βπΎ πΈπ΅πΈ πΏ = 0,27 ∗ + 0,32 ∗ π’ − 0,19 ∗ − 0,12 ∗ π − 0,04 ∗ πππ πΎ−1 ππ πΎ−1 ππ πΎ (1) The first dummy variable used in the model, for the profit rate equation, is introduced to account for the redistributive (or “rigorous”) measures imposed by the then weakened leftist government which were most evidently effective in the third quarter of 1986, when the rate of profit, after gradually increasing since 1983 from its former level of around 15%, went from 19 to 21% and has remained at around that level ever since, despite the 1990 and 2001 downturns. Therefore, a first statistically significant dummy is for the period 1986(3). A second dummy was used to account for another external shock, this time in capacity utilization and, as a consequence, in accumulation throughout the recent crisis. In the first quarter of 2008 accumulation reached 3.8% (more or less the same level reached before 1991 and 2001) then starts its steep downward path, which in the fourth quarter of the same year both capital accumulation and capacity utilization fell even more steeply, from 3.4 to 2.7% and from 86.8 to 85.5%, respectively, in order to fall even more (to 0.6% and 82.5%) in the first quarter or 2009. Therefore, another system-wide statistically significant dummy with one in the periods 2008(4) and 2009(1) and zeroes otherwise, was introduced. Needless to mention, these dummies are useful to correct for non-normality in the vectors of error terms. 43 Graph IV.1 (below) shows the series used in this first model along with their fitted values. The latter can barely be distinguished from the former, since the adjustment is surprisingly good. Graph IV.1 A brief economic explanation of our econometric findings runs as follows. Under a Kaleckian framework, higher profit rates and good demand conditions (here represented by utilization) positively influence capital accumulation, given that the former finance investment and the latter are necessary to keep sales flowing. Indebtedness in the private sector may have an initial positive effect on the economy, given the superior amount of resources it brings about. However, and as our results confirm, when debt increases more than proportionately than the stock of capital this brings about lower rates of accumulation by increasing the size of firm’s liabilities. The cost of credit, as measured by the interest rate, must be kept at low levels in order to promote investment. Finally, a high return on equities held has a negative effect on 44 fixed capital accumulation because this makes the financial sector grow at the expense of the real sector. This last point contrasts starkly with the neoclassical view in which financial accumulation (and, at an international level, integration) leads to higher levels of economic activity. IV.2 Debt ratio The second estimated model is a VAR(6) with an unrestricted constant relating the debt ratio with the rate of profit, the rate of return on equities issued and the interest rate. This long-run relationship reads as follows: πΏ πΈπ΅πΈ = 0,804 ∗ + 1,15 ∗ ππ − 7,73 ∗ π ππ πΎ ππ πΎ−1 (2) Graph IV.2 The sample runs for the same period as the previous model, and accounts for 108 parameters (6 lags, 4 variables, a constant term and two dummy variables) summarized in equation (2). Since the profit rate is one of the determinants of this system of equations, we include the 45 same 1986(3) dummy as in the previous model, which is statistically significant and may be interpreted along the same lines as before. The second dummy variable was used to account for a drastic change, which is not accounted for endogenously, in the rate of financial return in the second quarter of 1988. Immediately after the stock-market crash of 1987, financial profitability increases sharply from 8.3% in the first quarter of 1988 to 13% the period after. These are, as said before, the only external shocks which are not explained by our model and must be taken into account to ensure that the overall model is normal, homoskedastic and non-auto correlated. Graph IV.2 shows the series and their fitted values used in this model. This equation states that increases in the profit rate and in the rate of return on equities issued encourage indebtedness, whereas a higher interest rate depresses it, just as we mentioned in the accumulation function. The fact that the profit rate brings about increases in both investment and in indebtedness, and the latter depresses accumulation, is one of the main sources of cyclicality in our Minsky-inspired framework. Moreover, another source of cycles is that the financial rate of return depresses accumulation, though it increases the debt ratio. IV.3 Equity Issuing The following estimation corresponds to what we call an equity issuing equation, which is the stock of issued equities (peE) divided by total accumulation, both fixed (pkK) and financial (peEe). It comes from a VAR(5) with an unrestricted constant and four dummy variables, therefore it consists of 115 observations (the sample starts in 1981(2)) and, given that it contains 4 variables, 100 parameters. Equation (3) states that equity issuing depends positively on the interest rate and indebtedness, and negatively on the financial rate of return. πππΈ πΏ = 4,07 ∗ π + 1,04 ∗ − 2,01 ∗ πππ ππ πΎ + πππΈπ ππ πΎ (3) The first dummy is significant for three equations in the system: equity issuing, the financial rate of return and the interest rate, and the period is 1988(2). It was also used in the previous model to take into account an outlier in that period which occurred right after the stockmarket crash. Just one quarter before, the real interest rate reached its highest level from the sample (6.8%) and the share of equity issuing on accumulation fell from 64 to 62.7%. The latter did not represent the most drastic fall of the sample, but it is still significant and not accounted for by the series included in the model, or by their lags. 46 A second dummy was useful for two equations: equity issuing and the rate of financial return. This was effective in the second quarter of 1998, period in which financial profitability fell from 17 to 14.7% and equity issuing went from 84.3 to 89.2%. Another exogenous shock occurred in the second quarter of 2000, when equity issuing reached a peak (as high as 1.1), post stock-market boom hangover from which it has not yet “recovered”. Finally, in order to account for the drastic changes from the beginning of the eighties in the interest rate, we also included a dummy which is equal to -1 in 1981(3) (when r increased from 2.4 to 2.8%) and 1 in 1982(3) (when it went from 3.5 to 4.2%). Graph IV.3 shows the corresponding graphs. Graph IV.3 In economic terms, this long-run relationship states that increases in the interest rate discourage the demand for credit, thus it makes equity issuing more attractive to finance investment. Debt and equity issuing represent financing alternatives (external and internal, respectively) 47 but they are complementary rather than competing, as the positive sign of the debt ratio shows. Finally, financial profitability, via capital gains, makes financial wealth increase and the issuing of equities fall, thus the denominator increases faster than the numerator in the left-hand side of equation (3) as a consequence of an increase in the financial rate of return. IV.4 Financial Accumulation We now turn to one of the least studied equations in the macro-finance literature: financial accumulation. Our inference comes from a VAR(5) with an unrestricted constant and 6 dummy variables, thus a total of 112 estimated parameters. ππβπΈπ πΈπ΅πΈ πΏ = 0,054 ∗ + 0,015 ∗ πππ + 0,018 ∗ − 0,026 ∗ π (πππΈπ)−1 ππ πΎ−1 ππ πΎ (4) Equation (4) states that the growth rate of equities held depends positively on their rate of return and on the profit rate, whereas it depends negatively on the interest rate. As we are dealing with financial series, and these are well known to have an undesirably high volatile and speculative behavior (especially the variable of interest), we included several dichotomous variables which account for what our model cannot, leaving us with a statistically reliable long-run relationship. Thus, a first dummy is for the debt ratio10, which in 1981 experienced a sudden decline from 0.73 in 1980(4) to 0.71 the quarter after, to then fall again to 0.7. This dummy is then equal to -1 in 1981(1) and 1 in 1981(2). A second dummy is for the periods 1986(3) and 1987(3) when both financial accumulation and the profit rate suffered drastic changes, due to the stock-market crash. The period 1986(3), as explained in the accumulation and debt ratio equations, represents the ‘change of regime’ from “low” to high rates of profit. On the other hand, in the second quarter of 1987 the equity issuing ratio was 0.66, whereas in the following quarter it decreased to 0.65 and remained there the following three-month period, in order to continue its downward fall throughout 1988. We put these two dummies together in order to save degrees of freedom. The profit rate reaches a peak of 27% in the third quarter of 1998 (as a consequence of the peak in the rate of return and the debt ratio), and then starts a somewhat drastic fall 10 Debt is here divided by total capital, instead of productive capital. 48 throughout the following year11 and in the second quarter of 1999 it somewhat settles at a level around 25%. This dummy is then for 1999(2) which, again, is not accounted for by the variables and lags in the model. This also explains why this dummy did not appear in the previous models. This same dummy was significant in accounting for an outlier in the debt ratio equation at that same period. Graph IV.4 Finally, there are two structural changes which are also statistically and, much more important, economically significant. The first one goes from 1991 to the end of the sample, and it explains the shift in financial accumulation whose mean revolves around 1.15% before the shift, and around 0.94% right after. The second one is for the stock-market boom, which spans from 1995 to the first quarter of 1998, when the rate of return of equities held and the 11 However, by no means as important as to return to its former 15% level at the beginning of the eighties. 49 debt ratio started declining. Otherwise stated, when the bubble was uncovered. Of course, this dummy was also strongly significant for the rate of financial return. Appendix The sources of the data used are presented in Table A.2 below. Since data from the Comptes de Patrimoine provided by INSEE are presented on a yearly basis, and in some cases the same information is not available in their equivalent quarterly tables from Banque de France, we use the Denton (1971) technique to adjust annual data to quarters, ensuring that the sum of the values of all four quarters sums up to the annual figures from INSEE. This procedure was used for investment and the capital stock for French non-financial firms. Investment is calculated as the flow of Non-Financial Assets (Actifs non-Financiers) less depreciation (Consommation de Capital Fixe); Δ(pkK1). A second specification of investment also excludes Logements, Stocks and Actifs non Produits; Δ(pkK2). In both cases the quarterly series to which these annual figures were adjusted was Gross Fixed Capital Formation in constant terms. The capital stock was either defined as the stock of Actifs non-Financiers; pkK1, or as Actifs non-Financiers less Logements, Stocks and Actifs non produits; pkK2. The quarterly series used for the adjustment of these series were Δ(pkK1) and Δ(pkK2), respectively. All quarterly stock financial series come from Banque de France, and for these to be coherent with the data from the comptes de patrimoine from INSEE we carried out a method of étalonnage and calage described by INSEE (2011). Banque de France does not use either method when it presents the quarterly encours data. Instead, they carry out an exponential smoothing technique which does not correct for seasonal behavior. Therefore, we carry out the three methods as described by INSEE (2011). 50 A.I Computation of Fixed Capital Accumulation Based on Godley and Lavoie (2006) and dropping time subscripts, we know that the change in the capital stock (or investment) may be written as: β(ππ πΎ) = ππ πΎ − ππ−1 πΎ−1 Where Δ is a first difference operator, pk is the price of capital and K is the capital stock. This, in turn, can be rewritten as ππ πΎ + ππ πΎ−1 − ππ πΎ−1 − ππ−1 πΎ−1 ππ (πΎ − πΎ−1 ) + πΎ−1 (ππ − ππ−1 ) And rewriting K – K–1 as the difference between newly invested capital and capital depreciation, we have that we can further rewrite this as ππ (πΌ − πΏπΎ−1 ) + πΎ−1 βππ Where the termππ (πΌ − πΏπΎ−1 ) is the value of investment, and πΎ−1 βππ is the reevaluation effect. In order to calculate the volume of accumulation we divide ππ (πΌ − πΏπΎ−1 ) by ππ−1 πΎ−1, where the result is ππ (πΌ − πΏπΎ−1 ) ππ βπΎ βππ βπΎ = = (1 + ) ππ−1 πΎ−1 ππ−1 πΎ−1 ππ−1 πΎ−1 Therefore, the volume of accumulation must be corrected from changes in the price of capital, which is what we do using the growth rate of the price index of Gross Fixed Capital Forβππ mation as π π−1 . A.II Computation of Financial Accumulation In a similar way, we know that the change in the value of equities held is β(πππΈπ) = πππΈπ − ππ−1 πΈπ−1 51 Where pe is the price of equities and Ee is the volume of equities held. We rewrite the expression as πππΈπ + πππΈπ−1 − πππΈπ−1 − ππ−1 πΈπ−1 ππ(πΈπ − πΈπ−1 ) + πΈπ−1 (ππ − ππ−1 ) ππβπΈπ + πΈπ−1 βππ Dividing both sides of the equation by ππ−1 πΈπ−1 we obtain β(πππΈπ) ππβπΈπ βππ = + ππ−1 πΈπ−1 ππ−1 πΈπ−1 ππ−1 = (1 + We may call ππβπΈπ ππ−1 πΈπ−1 βππ βπΈπ βππ ) + ππ−1 πΈπ−1 ππ−1 the financial accumulation rate, which is the variable of interest in the financial accumulation equation. A.III Accounting Framework Table A.1 Actifs (A) Passifs (P) Actifs non Financiers (pkK1) Autres Actifs non Financiers (AANF) Passifs Financiers (PF) = Actifs Financiers (AF) = Crédit (L1 ou PF.4) Actifs Monétaires (M) + Autres Passifs (AP) + Autres Actifs (AA) + Actions Emises (peE ou PF.5) + Actions retenues (peEe ou AF.5) Net Wealth (NW = A – PF) 52 A.IV Data Sources Table A.2 Nomencl. Var. name Source INSEE. 8.212 Comptes de patrimoine et de variations de patrimoine des Δ(pkK1) Investment sociétés non financières (S11). AN ; Flux and CCF. Δ(pkK2) Investment less INSEE. 8.212 Comptes de patrimoine et de variations de patrimoine des logements, stocks and sociétés non financières (S11). actifs non produits AN, AN.1111, AN.12, AN.2 ; Flux and CCF. INSEE. 8.212 Comptes de patrimoine et de variations de patrimoine des pkK1 Capital stock sociétés non financières (S11). AN ; Stock. Capital stock less pkK2 INSEE. 8.212 Comptes de patrimoine et de variations de patrimoine des logements, stocks and sociétés non financières (S11). actifs non produits AN, AN.1111, AN.12, AN.2 ; Stock. Banque de France. Comptes Nationaux Financiers trimestriels, En- AF Financial Assets cours. Sum of all financial assets. Rows MU.Q.A.110000.212201 to MU.Q.A.110000.792000 FBCF_r FBCF A Gross Fixed Capital INSEE. Investissement des entreprises non financières - Biens - Volume Formation (constant) aux prix de l'année précédente chaînés. Gross Fixed Capital INSEE. Investissement des entreprises non financières - Biens - Valeur Formation (current) aux prix courants. Total Assets A = pkK1 + AF Banque de France. Comptes Nationaux Financiers trimestriels, En- PF Total Liabilities cours. Sum of all financial liabilities. Rows MU.Q.P.110000.292000 to MU.Q.P.110000.792000 peE Issued Equities (PF.5) Banque de France. Comptes Nationaux Financiers trimestriels, Encours. 53 MU.Q.P.110000.511100 + MU.Q.P.110000.512100 Banque de France. Comptes Nationaux Financiers trimestriels, EnpeEe Equities Held (AF.5) cours. Sum of rows MU.Q.A.110000.511100 to MU.Q.A.110000.520000 Banque de France. Comptes Nationaux Financiers trimestriels, En- L1 Loans (PF.4) cours. Sum of rows MU.Q.P.110000.411100 to MU.Q.P.110000.480000 Banque de France. Comptes Nationaux Financiers trimestriels, En- L2 Loans (PF.4 + PF.7) cours. PF.4 + row MU.Q.P.110000.792000 L3 Loans (PF – PF.5) L3 = PF – peE EBE Profit INSEE. Comptes des Sociétés non financières. Compte d’exploitation. NW Net Wealth NW = A – PF OF Own Funds OF = NW + peE INSEE. Comptes des Sociétés non financières. DD Distributed Dividends Comptes d’affectation des revenus primaires INSEE. Comptes des Sociétés non financières. RD Received Dividends Comptes d’affectation des revenus primaires Real long term Ecowin. Nominal interest rate and consumer price index. R interest rate All monetary data in current Euros, except FBCF_r. A.V Misspecification Tests Vector Misspecification Tests P-values The null hypotheses for the tests are: non-autocorrelation, normality and homoskedasticity. Thus, as all p-values lie above the 5% bound, we accept the null hypothesis, confirming our model is reliable on a statistical basis. The same applies for single equation tests, below. 54 Table A.3 Vector AR 1-5 test Vector Normality test Vector Hetero test F(180,238) χ2(12) χ2(1554) 0.0344* 0.2682 0.7263 F(80,254) χ2(8) F(480,314) 0.0982 0.3865 1.0000 F(80,266) χ2(8) F(400,417) 0.0155* 0.1503 0.9105 F(80,254) χ2(8) F(420,370) 0.0926 0.5452 1.0000 Accumulation Debt ratio Equity issuing Financial accumulation Single Equation Misspecification Tests P-Values Table A.4 Fixed Capital Accumulation Function Normality test ARCH 1-4 test Hetero test Accumulation 0.4133 0.7106 0.7311 Profit rate 0.1996 0.2411 0.3511 Utilization 0.1455 0.2113 0.5622 Debt ratio 0.7169 0.8327 0.2973 Interest rate 0.9041 0.5108 0.5029 Rate of financial return 0.0885 0.8782 0.8153 55 Table A.5 Indebtedness Function Normality test ARCH 1-4 test Hetero test Debt ratio 0.3322 0.7882 0.6643 Profit rate 0.6238 0.7133 0.9956 Rate of financial return 0.1523 0.7908 0.9979 Interest rate 0.5581 0.5526 0.8982 Table A.6 Equity Issuing Normality test ARCH 1-4 test Hetero test Equity issuing 0.4628 0.0268 0.3979 Interest rate 0.7751 0.5846 0.5399 Debt ratio 0.8796 0.6723 0.1688 Rate of financial return 0.0742 0.0624 0.9794 Table A.7 Financial accumulation Normality test ARCH 1-4 test Hetero test Financial accumulation 0.2397 0.2517 0.7620 Profit rate 0.7793 0.7450 0.9886 Rate of financial return 0.2943 0.8914 0.8559 56 0.7979 Debt ratio 0.6866 0.9943 Unit root tests The number of lags and the inclusion of constant and/or trend is based on t-tests. When the MacKinnon statistic is greater (in absolute value) than the critical value, we reject the null hypothesis of the existence of a unit root. Table A.8 shows that all series used in the models are not stationary in level. Table A.9 shows the tests for the variables in first difference and, as these are stationary, all series are integrated of order 1 or I(1). Table A.8 Series in level Characteristics Test Statistic / 1% Critical value 3 lags + constant – 2.947 / – 3.505 Accumulation 2 2 lags + constant – 3.202 / – 3.504 Utilization 3 lags – 1.989 / – 2.598 Profit Rate 1 2 lags – 0.468 / – 2.598 Profit Rate 2 2 lags + constant – 1.696 / – 3.504 L1/K1 2 lags – 0.036 / – 2.598 L2/K1 2 lags – 0.070 / – 2.598 L3/K1 2 lags – 0.390 / – 2.598 L1/K2 2 lags + cnst + tr – 2.551 / – 4.034 L2/K2 2 lags + cnst + tr – 2.874 / – 4.034 L3/K2 2 lags + cnst + tr – 2.323 / – 3.488 Interest rate 2 lags + constant – 2.641 / – 3.504 Variable Accumulation 57 Table A.9 Series in differences Variable Characteristics Test Statistic / 1% Critical value Δ(Accumulation) 2 lags – 5.644 / – 2.598 Δ(Accumulation 2) 1 lag – 6.210 / – 2.598 Δ(Utilization) 2 lags – 6.688 / – 2.598 Δ(Profit Rate 1) 1 lag – 5.891 / – 2.598 Δ(Profit Rate 2) 1 lag – 5.808 / – 2.598 Δ(L1/K1) 1 lag – 7.632 / – 2.598 Δ(L2/K1) 1 lag – 7.560 / – 2.598 Δ(L3/K1) 1 lag – 7.580 / – 2.598 Δ(L1/K2) 1 lag + constant – 8.458 / – 3.504 Δ(L2/K2) 1 lag + constant – 8.433 / – 3.504 Δ(L3/K2) 1 lag + constant – 8.637 / – 2.889 Δ(Interest rate) 1 lag + constant – 6.688 / – 2.598 Lags tests As these are exclusion tests, the null hypothesis is that the corresponding set of coefficients (that is, for all equations in the system) is equal to zero. Thus, for p-values less than 0.05 the lag is statistically significant at a 5% level. Table A.10 P-values per lag in Accumulation function Varuiable \ Lag 1 2 3 4 5 6 Accumulation 0.000 0.460 0.056 0.771 0.676 0.426 Profit rate 0.000 0.387 0.080 0.075 0.302 0.049 58 Utilization 0.000 0.000 0.043 0.032 0.001 0.004 Debt ratio 0.000 0.027 0.895 0.804 0.105 0.006 Interest rate 0.000 0.013 0.830 0.171 0.015 0.526 Rate of financial return 0.000 0.000 0.265 0.001 0.000 0.017 Table A.11 P-values per lag in Debt ratio function Varuiable \ Lag 1 2 3 4 5 6 Debt ratio 0.000 0.003 0.654 0.985 0.373 0.203 Profit rate 0.000 0.275 0.042 0.024 0.514 0.411 Interest rate 0.000 0.001 0.720 0.029 0.000 0.039 Rate of financial return 0.000 0.000 0.262 0.000 0.000 0.001 Table A.11 P-values per lag in Equity issuing function Varuiable \ Lag 1 2 3 4 5 Equity issuing 0.000 0.002 0.169 0.000 0.000 Interest rate 0.000 0.009 0.916 0.282 0.002 Debt ratio 0.000 0.008 0.809 0.792 0.657 Rate of financial return 0.000 0.000 0.135 0.543 0.012 59 Table A.11 P-values per lag in Financial accumulation function Varuiable \ Lag 1 2 3 4 5 Financial accumulation 0.000 0.004 0.621 0.720 0.451 Profit rate 0.000 0.162 0.182 0.115 0.299 Rate of financial return 0.000 0.003 0.256 0.000 0.000 Debt ratio 0.000 0.000 0.767 0.958 0.234 References Aglietta, M. (1998), ‘Le Capitalisme de demain’, Notes de la fondation Saint-Simon, n° 101. Boyer R. (2000): ‘Is a finance-led growth regime a viable alternative to Fordism? A preliminary analysis’, Economy and Society, 29 (1), pp. 111-145. Brainard, W.C. and J. Tobin (1968) ‘Pitfalls in financial model building’, American Economic Review, 58 (2) (May), pp. 99-122. Clévenot, M., Y. Guy and J. Mazier (2010) ‘Equity and debt in a financialized economy: the French case’ CEPN working paper. Dimand, R.W. (2003) ‘James Tobin and the Transformation of the IS/LM Model’. Presented to the History of Political Economy conference on “The IS/LM Model: Its Rise, Fall, and Strange Persistence” Duke University. Denton, Frank T. “Adjustment of Monthly or Quarterly Series to Annual Totals: An Approach Based on Quadratic Minimization”. Journal of the American Statistical Association. March 1971, Volume 66, Number 333. Dos Santos, C.H. (2005) ‘A stock-flow consistent general framework for formal Minskyan analyses of closed economies’, Journal of Post Keynesian Economics 27 (4) (Summer), pp. 711-36. 60 Duwicquet V. and J. Mazier (2011) ‘Financial integration and macroeconomic adjustments in a monetary union’, Journal of Post Keynesian Economics, vol. 33, no. 2, pp. 331368. Galbraith, J.K (1967), The New Industrial State, Boston. Houghton Mifflin. Godley, W. and M. Lavoie (2007) Monetary Economics: An Integrated Approach to Credit, Money, Income, Production and Wealth. Palgrave Macmillan. Guy, Y., M. Clévenot, and J. Mazier (2010) ‘Investment and the rate of profit in a financial context: the French case’, International Review of Applied Economics. INSEE, (2011). Méthodologie des comptes trimestriels – Chapitre 2. http://www.insee.fr/fr/ppp/sommaire/imet108b.pdf Lavoie, M. and W. Godley (2001) ‘Kaleckian Growth Models in a Stock and Flow Monetary Framework: A Kaldorian View’, Journal of Post Keynesian Economics, 24 (2) (Winter), pp. 277-312. Minsky H. (1986) ‘Stabilizing an Unstable Economy’, McGraw Hill. Stockhammer E. (2004): ‘Financialization and the slowdown of accumulation’, Cambridge Journal of Economics, 28 (5), pp. 719-742. Taylor, L. (2010) ‘Growth, cycles, asset prices and finance’ Metroeconomica. Tobin, J. (1969) ‘A general equilibrium approach to monetary theory’, Journal of Money, Credit, and Banking, 1 (1) (February), pp. 15-29. Zezza G. and C.H. Dos Santos (2004) ‘The role of monetary policy in post-Keynesian stock-flow consistent macroeconomic growth models’, in M. Lavoie and M. Seccareccia (eds), Central Banking in the Modern World: Alternative Perspectives (Cheltenham, UK, and Northampton, USA: Edward Elgar), pp. 183-208. 61 Initial values of parame- g0 = 0.2352693030… K = 400 ters g1 = 0.3 L = 100 g2 = 0.04 pe = 35 g3 = 0 rl = 0.02 z0 = 0.3 r = 0.02 z1 = 0.5 rb = 0.02 z2 = 0.45 TB = 0.393063 z3 = 0.033333… TCB = 0.176982075 θ = 0.1 a0 = 0.5658628 a1 = 0.83 a2 = 0.04 v0 = 0.22382378 T = 7.47687 v1 = 0.2 v2 = 0.2 Values for exogenous variables v3 = 0.1 ib = 0.015 w0 = 0.389734150 id = 0.01 UP = 23.6813 Vh = 89.54858 YHSh = 67.2918 YDh = 67.2918 w1 = 0.01 Y = 100 w2 = 0.02 Initial values w3 = 0.02 BD = 45 λ0 = 0.159143 B=0 k0 = 0.1086334242… BP = 0.979955 k1 = 0.35 BT = 0 k2 = 0.025 C = 60 k3 = 0.1 DIV = 20 k4 = 0.5 DIVe = 13.33... k5 = 0.1 DIVh = 6.66... δ = 0.0625 Vg = 0 f0 = 0.09826265506 E=3 f1 = 0.2 Ee = 2 f2 = 0.6 Eh = 1 r0 = 0.67652 g = 0.0625 sf = 0.34097798866 Hh = 9.54858 θb = 0.2862767 Hb = 2.250225 λ = 0.050005 I = 25 m1b = 0.005 m2b = 0.005 W = 67.652 H = 11.798805 RF = 11.798805 G = 15 pb = 50 62