Credit management has important effects on the availability of investible funds for the private sector; it is quite often the cornerstone of financial reforms aimed at increasing investment, and hence, industrial growth rates. In this paper, we examine the potential impact of interest rate policies on investment in the mining and manufacturing sector of Greece.

Financial markets in Greece are not well developed. The virtual nonexistence of markets for bonds and equity results in a situation where the private sector depends on bank credit as its only source of external finance. Bank credit is heavily subsidized, so that both deposit and lending rates set by the Bank of Greece have been predominantly negative in real terms since 1973.

As a result, one would expect credit availability to be restricted, and private industrial investment to be constrained, by the shortage of funds. However, survey data from firms in the private manufacturing sector, as reported in Deleau (1987), do not support this expectation. Most individual firms surveyed do not perceive a shortage of funds as restricting investment. In effect, even negative real interest rates do not provide an effective subsidy. This would appear to be somewhat of a puzzle. The “naive” explanation implies that the rate of return on capital has been persistently negative across firms and types of productive activity; but this explanation is a difficult one to maintain consistently.

We argue here that the puzzle arises only if we think of bank credit as the sole source of long-term finance. Consider a firm that operates in a credit-controlled environment; in addition to borrowing, the firm can finance its borrowing with corporate savings (retained profits). If borrowing is rationed repeatedly, the firm will use profit retention as a source of finance to optimize on its investment outlay. We should visualize this firm as optimizing jointly over its investment budget and profit-retention rate, taking into account the credit ceiling and debt-servicing costs.

The model of credit-constrained investment financing has rather different implications than does a more naive model where firms react passively by setting equate their investment budgets with the level of borrowing. The savings function, s, determines the amount banks can lend to firms; and *i* is the ex-ante investment function. The ex-post investment function î is the minimum of exante savings and investment. At low interest rates, î_{r} > 0.

Figure 2 demonstrates the ex-post investment function of firms that choose financing methods optimally, relative to a credit constraint. Household savings, *sh*, is the amount that banks can lend. Corporate savings out of profits is a function, *s*π, which is unambiguously decreasing in *r*. Investment, both ex ante and ex post, equals *i* = *sh* + sπ. It follows that *i*_{r}, < 0 whenever *sh*_{r} + sπ < 0. Our point estimates suggest that this is true for investments in the Greek manufacturing sector, even for real interest rates as low as -10 percent.

For our argument to be empirically convincing, we need to establish that corporate savings are at once a major and a discretionary component of investment finance. Figures 3, 4, and 5 illustrate some evidence to this effect. In Figure 5, we notice, first, that the proportion of investment financed by retained earnings has varied substantially from above 80 percent in the mid-1960s to below 50 percent in the mid-1980s. In Figures 3 and 4, we see that the time path of corporate savings displays properties very similar to that of investment: this apparent parallel movement points to its importance as a discretionary source. An important indicator, apparent in Figure 5, is the long-term decline in the role of corporate savings in financing investments. Others, including Tsoris (1984) and Deleau (1987), have noted that the indebtedness of Greek manufacturing firms is high and increasing. Our argument associates this rise with declining profitability and consequent cutbacks in real investment outlay.

**Greece: Investment and Corporate Savings—Mining and Manufacturing, 1964–86**

(In billions of Greek drachmas)

**Greece: Investment and Corporate Savings—Mining and Manufacturing, 1964–86**

(In billions of Greek drachmas)

**Greece: Investment and Corporate Savings—Mining and Manufacturing, 1964–86**

(In billions of Greek drachmas)

**Greece: Real Investment and Corporate Savings (1970 Prices), 1964–86**

(In billions of Greek drachmas)

**Greece: Real Investment and Corporate Savings (1970 Prices), 1964–86**

(In billions of Greek drachmas)

**Greece: Real Investment and Corporate Savings (1970 Prices), 1964–86**

(In billions of Greek drachmas)

**Greece: Corporate Savings as a Percentage of Investment, 1964–86**

**Greece: Corporate Savings as a Percentage of Investment, 1964–86**

**Greece: Corporate Savings as a Percentage of Investment, 1964–86**

The empirical analysis suggests that one of the major determinants of this decline in profitability is the long-term increase in product-wage rates in manufacturing, a natural outcome of government policy measures aimed at increasing workers’ standard of living. The questions that should be addressed in this context deals with the design of financial reforms in the presence of redistributive measures.

A related set of questions, which we do not address here, deals with the role of interest rate subsidies in the nonexistence of securities markets, which, to a great extent, must be traced to institutional factors, including the legal framework supporting the rights of investors. However, the availability of subsidized credit can support and perpetuate the existing structure of “family-based” enterprises, which do not need to dilute control in order to increase their capital base with equity finance.

We think that questions about the lack of equity financing are important precisely because of redistributive policy. If equity were a predominant source of finance, real wage increases would generate very different spillover effects on the investment activities of firms. A policy aimed at taking advantage of these effects must focus financial reforms on securities markets, rather than debt markets, which appear to be of limited effectiveness anyway.

## I. Modeling the Investment Decision

### The Firm—Credit Constraints

We consider a firm that has a given technology, using capital, *k*, and labor, *ℓ*, to produce an output, *x*. Each period, the firm faces a product price, a wage rate, and a price for the investment good. Investment decisions have to be made one period in advance, since capital needs time to build. Each period, the firm starts with a given capital stock; it decides on its level of employment and, hence, on total output; in addition, it has to decide on the amount of investment, *i*, so that next period’s capital stock is *k*’ = *i* + (1 - δ)*h*, with δ as the one-period depreciation rate for capital.

Investment is financed by one of two means: long-term borrowing from the banking sector; or retained profits by the firm (corporate savings). We will assume, for simplicity, that all borrowing in period *t* has to be fully repaid in period *t* + 1, so that the repayment period equals the time required for building productive capacity. We focus on a situation where interest rates are managed, so that real interest rates are low relative to the possible rate of return on capital; this implies that at the announced rate of interest, the firm is constrained in its borrowing, and can only borrow an amount b, a credit ceiling set by the bank; if it wishes to invest more than this, it must finance the additional amount from retained earnings.

The firm is owned by shareholders, who are paid dividends; profits of the firm are either retained to finance new investment or paid out as dividends. In particular, we abstract from working capital needs, as well as constraints on short-term borrowing. In each period, the shareholders have to decide simultaneously on the dividend payout rate and the level of investment, taking into account the credit ceiling faced by the firm. In fact, since the bank informs the firm about its borrowing limits, investment and dividend payouts are outcomes of the same decision.

The following identities summarize the basic structure of the problem:

where the product price is normalized to equal 1; *q* is the relative price of investment goods; *w* is the real wage rate; *r* is the real (long-term lending) rate charged by the bank; *b* is the credit ceiling; π, *d*, and ; are real profits, dividends, and investments, respectively. Variables determined in the previous period are denoted by an overbar (¯) and in the following period by a prime symbol (’).

The production function *f* is increasing and concave in both arguments, and displays constant returns to scale. We assume, further, that shareholders care only about real dividend payments, so that *x* can be thought of as a composite consumption good.

It may be useful to think of the firm’s decision as being taken in two steps; it begins a production period with a given capital stock. The firm, which is a price taker in the product and capital markets, first chooses a level of employment. In doing so, it maximizes profits, so that the marginal productivity of labor is equated to the product wage rate. Once profits are realized, the firm has to make a second decision—to split profits into dividends and corporate savings. At this point, the firm needs to know, in addition, the credit ceiling set by the bank, and to anticipate the real wages and the real interest rate it will face next period; once this is known, the shareholders know the (expected) contribution of retained earnings to profits next period, so that they choose to trade off dividends today for profits tomorrow. In the following section, we indicate the nature of this optimization problem and the determinants of investment financing.

### Optimization and Dividend Decisions

A representative shareholder derives utility from dividend income; the shareholder’s objective function is summarized as a two-period utility function:

where both *u* and *v* are increasing and concave in their respective arguments. The two-period representation can be justified either as representing shareholder myopia, or as a representation of a longer-term optimization problem via the value function.

Since the firm operates by maximizing profits each period, it must be true that

which we can rewrite as

with *h*_{k} > 0; *h*_{w} < 0. The function *h* is simply the potential operating profits achievable by the firm with a given capital stock and facing a given real wage rate in the labor market. It follows that realized profits are operating profits minus effective debt-servicing costs:

The budget constraint facing the representative shareholder is thus defined by equation (8), and, in addition

The firm now chooses its payout policy to maximize equation (5), subject to equations (8) and (9); the optimal payout rule can be written either in terms of dividends or in terms of real retained profits, which is just the residual. Thus, optimal retained profits will be

Notice that in order to implement this decision, the firm has to know the product wage rate next period, the inflation rate for its own output, and the inflation rate on capital goods p’ = (q’/q) -1; in the actual estimation procedure, we use both the “perfect foresight” as well as a “rational expectations” specification to examine the responsiveness of investment decisions.

### Comparative Statics

The function *h* defines potential operating profits at a given capital stock and wage rate; from the linear homogeneity (constant returns to scale) of the production function, it follows that the rate of return on capital is constant for a given wage rate.

Writing this return as *r*(*w*) a decreasing function of *w*, we have

Realized profits next period are then

The effective return on investment expenditure each period is

It follows that for *r** < *r*, if the effective rate of return on investment is less than the real interest rate, the firm will disinvest and, in particular, will not borrow; since our interest is to model a situation where the interest rate subsidy is effective and the credit ceiling does in fact bind, we assume that

A shareholder thus maximizes

and the first-order conditions are

Totally differentiating both sides, we obtain

with τ ≡ *u*″/*v*″ > 0. It follows that

The firm’s propensity to save, *s*_{1}, is less than 1. The negative impact of increased credit availability on corporate savings can be understood as follows: if the real rate of return is relatively constant, and higher than the unit cost of debt servicing, firms would like to borrow to finance their entire investment plans. To the extent they cannot, corporate savings finance the gap; whenever they are allowed to borrow more, they can use current profits to finance a higher dividend payout Notice that *s*_{2} < 1, as long as *r* >-1; and further, that *s*_{2} > 0 if and only if *r** > *r*, as we have in fact assumed. The hypothesis *r** = *r* (firms are not constrained in their borrowing) is thus easily testable, either with aggregate data or with cross-sectional analysis.

The coefficient *s*_{2} indicates the extent to which bank credit in fact subsidizes investment. Notice that for τ = 0,

which gives us a point estimate of the markup of profitability over the interest factor.

## II. Data Analysis and Empirical Evidence

In the last section, we modeled the internal decision making by a firm that determines the corporate savings component of its investment; the firm takes into account the fact that it can only borrow up to a preset credit limit and then determines its total investment budget as borrowing plus own savings. The fact that it is constrained in borrowing alters the nature of its optimal investment. We showed that corporate savings are a decreasing function of total debt-servicing costs and, hence, of the real interest rate and of the credit ceiling, even if the real interest rate is zero or negative.

### Empirical Implications

This model has important implications for the empirical analysis of the investment decision. In the simplest version of a constrained (or disequilibrium) investment model, the reduced form is derived as follows: total investible funds equal available savings, domestic and foreign, which is an increasing function of the real interest rate, as well as of variables such as disposable income, *y*; investment expenditure of the firm is a decreasing function of the real interest rate, and depends, in addition, on other factors affecting firm profitability, summarized as z. The equilibrium interest rate, *r*_{e} = *r*(*y, z*) equates savings to exante investment; if the government (or the banks), which manages the interest rate, offers an interest rate different from *r*_{e}, and in particular, below *r*_{e}, the capital market is in disequilibrium, and actual investment is constrained by the availability of savings. This can be written as

where *i* is the ex-ante level of investment, and î the actual, or ex-post level. Clearly, if *r*, the administered interest rate, is below *r*_{e}, actual investment is less than intended; and for this regime, the model predicts, empirically, that î is increasing in *r* and in *y*. Increasing the real interest rate will loosen the credit constraint on industry, and increase the availability of investible resources. This argument is a simplified version of the model familiar both from the literature on “fixed prices” (see, for example, Malinvaud (1980) and (1985)), as well as on interest rate management (see, for example, International Monetary Fund (1987), which cites the well-known McKinnon-Shaw argument).

This argument envisages, on the one hand, firms that formulate their investment decision a priori, and are then repeatedly faced with a credit ceiling. On the other hand, firms that operate in a perpetually credit-constrained environment are likely to take this ceiling into account when making their investment plans, and organize their investment and financing decisions on this basis. The model of the previous section describes the nature of the constrained investment decision. To write the comparable reduced form for this model, we decompose total private sector savings into two components: household savings, *sh*, and corporate savings, sπ. This distinction is important precisely because interest rates are managed, which drives a wedge between the return on debt—that is, the bank rate—and the return on capital. Household savings responds to *r*, the bank rate, as well as to personal disposable income, *y*. (Strictly speaking, *y* should include income from sources other than profits.) All household savings are held as bank deposits, which can be lent out to firms to finance investment; if reserve requirements are ignored for the moment, this represents the credit ceiling set by the banking sector as a whole. Firms finance investment with borrowing and with corporate savings; corporate savings depend on current profits, the credit ceiling, and the bank rate, as well as on factors *z*, which determine the rate of return on capital internal to the firm; ex-ante constrained investment is equal to available credit plus corporate savings:

When the credit constraint is binding—that is, *b* = *sh*, we can write the investment function in reduced form as

It follows that

Evidently, *i*_{r} can be negative; and is strictly decreasing in y. Denoting ε*(sh)* as the interest elasticity of household savings, and (*i*) as the interest elasticity of private investment, we have

with α the proportion of total private savings held by households; it can be seen that at *r* = 0, ε(*i*) is negative if

impact of a relaxation of credit ceilings. If this change is autonomous (for example, if it is funded by a net capital inflow), the net effect is to increase investment by a factor of (1-*s*_{2}) which can be estimated from a direct estimate of equation (18). However, if the change is induced by an increase in real interest rates, the net effect has to incorporate the effect of increased debt-servicing costs faced by the firm; and, as equation (23) shows, this is likely to reduce total investment. We estimate *s*π_{b(l+r)} and ε*(sh)*, and from there, the implied interest elasticity of investment.

### On Data and Trends

For the present, we concentrate on patterns of investment in the aggregate mining and manufacturing sector. All data are from the Organization for Economic Cooperation and Development’s *OECD Economic Surveys: Greece* (OECD (various years)), which, in turn, are computed from raw data provided by the Bank of Greece. One of the problems is that direct data for corporate savings (either levels or rates) are not available; from 1967 onwards, the National Income Accounts of Greece include business savings in the category of private sector and household savings—that is, the residual after consumption and public savings. According to the OECD, “[t]he relatively small decline in private savings seems to conceal two divergent trends. Household savings appear to have increased since 1980, whereas business savings have fallen markedly” (OECD (1986), p. 36).

We estimated the corporate savings component as the increase in nominal gross capital formation in mining and manufacturing, minus the increase of long-term lending by the banking and financial services sector; clearly, this estimation makes two implicit assumptions: first, that the financing of long-term investment projects is obtained either from the banks or from corporate savings (including any possible interfirm credit); and second, that all long-term lending is in fact spent on gross investment. There is some empirical basis for the first kind of assumption. Tables 1 and 2 below portray recent data on the financing of investment in the Greek economy; Table 1 gives the sources of financing for aggregate investment, and Table 2, the sources of external finance for the private sector.

**Greece: Investment and Its Financing, 1981–85**

(In percent of resources available for investment)

**Greece: Investment and Its Financing, 1981–85**

(In percent of resources available for investment)

Resource | 1981 | 1982 | 1983 | 1984 | 1985 |
---|---|---|---|---|---|

Net lending from abroad | 24.4 | 20.7 | 23.5 | 29.5 | 39.5 |

Private sector savings | 109.7 | 110.3 | 105.1 | 107.3 | 105.5 |

Public sector | −34.1 | −31.0 | −28.5 | −36.8 | −45.0 |

**Greece: Investment and Its Financing, 1981–85**

(In percent of resources available for investment)

Resource | 1981 | 1982 | 1983 | 1984 | 1985 |
---|---|---|---|---|---|

Net lending from abroad | 24.4 | 20.7 | 23.5 | 29.5 | 39.5 |

Private sector savings | 109.7 | 110.3 | 105.1 | 107.3 | 105.5 |

Public sector | −34.1 | −31.0 | −28.5 | −36.8 | −45.0 |

**Greece: Private Sector Financing, 1982–84**

(In percent of total)

**Greece: Private Sector Financing, 1982–84**

(In percent of total)

Resource | 1982 | 1983 | 1984 | |
---|---|---|---|---|

Bank credit | 86.9 | 80.2 | 79.3 | |

Short term | 46.0 | 38.9 | 42.1 | |

Long and medium term | 40.9 | 41.3 | 37.2 | |

Security issues, net | 00.2 | 00.3 | 00.7 | |

Net capital inflows from abroad | 12.9 | 19.5 | 20.0 | |

Real estate | 9.7 | 13.3 | 14.2 | |

Business | 3.1 | 5.7 | 4.4 | |

Other | 0.1 | 0.5 | 1.4 |

**Greece: Private Sector Financing, 1982–84**

(In percent of total)

Resource | 1982 | 1983 | 1984 | |
---|---|---|---|---|

Bank credit | 86.9 | 80.2 | 79.3 | |

Short term | 46.0 | 38.9 | 42.1 | |

Long and medium term | 40.9 | 41.3 | 37.2 | |

Security issues, net | 00.2 | 00.3 | 00.7 | |

Net capital inflows from abroad | 12.9 | 19.5 | 20.0 | |

Real estate | 9.7 | 13.3 | 14.2 | |

Business | 3.1 | 5.7 | 4.4 | |

Other | 0.1 | 0.5 | 1.4 |

Tables 1 and 2 demonstrate that the only source of financing available to private firms other than bank loans is foreign capital; we have ignored this component in calculating corporate savings. Apart from being of a small order of magnitude, some of this capital inflow is in the form of ownership, and likely to be closely approximated by a proportional factor of corporate savings.

Table 3 shows the average long-term lending rates for industry, the inflation rate in product prices, and the ex-post real interest rate. As evident from this table, real interest rates have been largely negative from 1972 on.

**Greece: Interest Rates, 1961–85 (In percent)**

**Greece: Interest Rates, 1961–85 (In percent)**

Lending | Forward | Real Interest | |
---|---|---|---|

Year | Rate | Inflation | Rate (ex post) |

1961 | 7.500000 | 1.458097 | 5.955072 |

1962 | 7.500000 | 1.676595 | 5.727380 |

1963 | 7.500000 | 1.766801 | 5.633663 |

1964 | 7.500000 | 3.009296 | 4.359513 |

1965 | 7.500000 | 3.483105 | 3.881692 |

1966 | 7.670000 | 1.628697 | 5.944485 |

1967 | 8.000000 | −0.213671 | 8.231258 |

1968 | 8.000000 | 2.569604 | 5.294352 |

1969 | 8.000000 | 4.384100 | 3.464033 |

1970 | 8.000000 | 1.800001 | 6.090372 |

1971 | 8.000000 | 3.339899 | 4.509489 |

1972 | 8.000000 | 18.251000 | −8.668848 |

1973 | 8.830000 | 22.588400 | −11.223250 |

1974 | 10.830000 | 9.114801 | 1.571921 |

1975 | 11.000000 | 12.860600 | −1.648583 |

1976 | 11.000000 | 11.235400 | −0.211624 |

1977 | 11.000000 | 8.568705 | 2.239407 |

1978 | 11.580000 | 18.165800 | −5.573356 |

1979 | 13.580000 | 24.328400 | −8.645168 |

1980 | 17.250000 | 22.959200 | −4.643166 |

1981 | 18.500000 | 18.965100 | −0.390954 |

1982 | 16.500000 | 21.091500 | −3.791760 |

1983 | 18.500000 | 20.162700 | −1.383708 |

1984 | 18.500000 | 18.104900 | 0.334533 |

1985 | 18.500000 | 25.250700 | −5.389751 |

**Greece: Interest Rates, 1961–85 (In percent)**

Lending | Forward | Real Interest | |
---|---|---|---|

Year | Rate | Inflation | Rate (ex post) |

1961 | 7.500000 | 1.458097 | 5.955072 |

1962 | 7.500000 | 1.676595 | 5.727380 |

1963 | 7.500000 | 1.766801 | 5.633663 |

1964 | 7.500000 | 3.009296 | 4.359513 |

1965 | 7.500000 | 3.483105 | 3.881692 |

1966 | 7.670000 | 1.628697 | 5.944485 |

1967 | 8.000000 | −0.213671 | 8.231258 |

1968 | 8.000000 | 2.569604 | 5.294352 |

1969 | 8.000000 | 4.384100 | 3.464033 |

1970 | 8.000000 | 1.800001 | 6.090372 |

1971 | 8.000000 | 3.339899 | 4.509489 |

1972 | 8.000000 | 18.251000 | −8.668848 |

1973 | 8.830000 | 22.588400 | −11.223250 |

1974 | 10.830000 | 9.114801 | 1.571921 |

1975 | 11.000000 | 12.860600 | −1.648583 |

1976 | 11.000000 | 11.235400 | −0.211624 |

1977 | 11.000000 | 8.568705 | 2.239407 |

1978 | 11.580000 | 18.165800 | −5.573356 |

1979 | 13.580000 | 24.328400 | −8.645168 |

1980 | 17.250000 | 22.959200 | −4.643166 |

1981 | 18.500000 | 18.965100 | −0.390954 |

1982 | 16.500000 | 21.091500 | −3.791760 |

1983 | 18.500000 | 20.162700 | −1.383708 |

1984 | 18.500000 | 18.104900 | 0.334533 |

1985 | 18.500000 | 25.250700 | −5.389751 |

Figures 3–6 show trends in the calculated corporate savings component over time, real and nominal and relative to investment and to revenue. The negative trend in the corporate savings rates is clearly discernible in Figures 5 and 6. Figures 7 and 8 show trends in variables that are exogenous to the investment decision, factor prices, and the real cost of borrowing.

**Greece: Corporate Savings as a Percentage of Revenue, 1964–86**

**Greece: Corporate Savings as a Percentage of Revenue, 1964–86**

**Greece: Corporate Savings as a Percentage of Revenue, 1964–86**

**Greece: Relative Price of Capital Goods**

(Index, 1960 = 1)

**Greece: Relative Price of Capital Goods**

(Index, 1960 = 1)

**Greece: Relative Price of Capital Goods**

(Index, 1960 = 1)

### Empirical Analysis

We estimate the parameters of the corporate savings function in two versions. In the first version, we estimate directly the linearized version of the reduced form specified by equation (10), or equivalently, equation (20b). This yields a point estimate of the parameter *s*_{b(l+r)};whereas direct estimation of equation (20) yields estimates for the interest elasticity of savings—that is, ε*(sh)*. We can use this to simulate the interest elasticity of investment expenditure at different real interest rates.

In the second version, we use the approximation suggested by equation (18). The long-run, or steady-state, corporate savings decision depends on profits and the availability of credit; the estimation yields a point estimate of *s*_{2}, and, hence, the long-run response of investment to changes in credit availability, as (1 - *s*_{2}); this also gives us an approximation of the interest subsidy factor. We then model the deviation from the steady-state corporate savings level as an adjustment process, affected by anticipations of the real wage rate and the interest rate. Even though constraints on the availability of data preclude more ambitious modeling at the moment, there is strong evidence that these deviations (or equilibrium errors) have some persistence over time, even though they dissipate in the long run. This means, for example, that a large increase in the real wage rate leads to a cutback in current investment expenditures, and this negative effect persists for a few periods, leading to investment cutbacks in subsequent periods as well.

Table 4 reports the results of the analysis of the corporate savings function. We used real revenues (that is, value added deflated by a product price index) to approximate profits, because of the unavailability of reliable data on sector-specific profits in the private sector. A first-order moving average term is incorporated to account for possible effects generated by capital accumulation, since real investment is endogenous. As it turns out, this is not statistically significant. Equation (25) estimates the relation with the assumption of perfect foresight; equation (26) reports the results of a “rational expectations” assumption, so that real wages and ex-post real interest rates are correctly anticipated on average, given current information. The estimation technique is instrumental variables (IV); the current real wage rate and real interest rate are used as instruments.

**Greece: The Corporate Savings Function—Estimates for 1962–86**

*rev*denotes post-tax revenues;

*b*denotes long-term borrowing from banks;

*w*’ is the (forward) index of manufacturing wages; q is the index of capital goods prices; MA is the one-period moving average;

*R*denotes the coefficient of determination, and D-W is the Durbin-Watson stastistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

^{2}**Greece: The Corporate Savings Function—Estimates for 1962–86**

Coefficients | ||||||||
---|---|---|---|---|---|---|---|---|

Corporate Savings | ||||||||

Function | c | rev | b(1+r) | w′ | q | MA | R^{2} | D-W |

Equation (25) | 0.1754 | 0.2570 | −0.4309 | −0-0486 | −0.148 | 0.18 | 0.87 | 1.94 |

OLS (t-ratios) | (3.95*) | (5.87*) | (1.89**) | (1.96**) | (3.39*) | (0.69) | ||

Equation (26) | 0.1433 | 0.3021 | −0.30B9 | −0.0820 | −0.1120 | 0.10 | 1.61 | 1.87 |

IV (t-ratios) | (1.48*) | (7.35*) | (2.33*) | (3.35*) | (1.06) | (1.09) |

*rev*denotes post-tax revenues;

*b*denotes long-term borrowing from banks;

*w*’ is the (forward) index of manufacturing wages; q is the index of capital goods prices; MA is the one-period moving average;

*R*denotes the coefficient of determination, and D-W is the Durbin-Watson stastistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

^{2}**Greece: The Corporate Savings Function—Estimates for 1962–86**

Coefficients | ||||||||
---|---|---|---|---|---|---|---|---|

Corporate Savings | ||||||||

Function | c | rev | b(1+r) | w′ | q | MA | R^{2} | D-W |

Equation (25) | 0.1754 | 0.2570 | −0.4309 | −0-0486 | −0.148 | 0.18 | 0.87 | 1.94 |

OLS (t-ratios) | (3.95*) | (5.87*) | (1.89**) | (1.96**) | (3.39*) | (0.69) | ||

Equation (26) | 0.1433 | 0.3021 | −0.30B9 | −0.0820 | −0.1120 | 0.10 | 1.61 | 1.87 |

IV (t-ratios) | (1.48*) | (7.35*) | (2.33*) | (3.35*) | (1.06) | (1.09) |

*rev*denotes post-tax revenues;

*b*denotes long-term borrowing from banks;

*w*’ is the (forward) index of manufacturing wages; q is the index of capital goods prices; MA is the one-period moving average;

*R*denotes the coefficient of determination, and D-W is the Durbin-Watson stastistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

^{2}Equation (25)—that is, the perfect foresight hypothesis—yields a point estimate of 0.4309 for *s*π_{b(l + r)}, whereas the rational expectations hypothesis (equation (26)) yields 0.3089. In both versions, initial estimation with ρ (capital goods inflation) showed that its coefficient was statistically insignificant from zero.

To complete this version, we estimate an equation representing the interest elasticity of credit availability; that is, the responsiveness of long-term lending by banks to the mining and manufacturing sector to variations in the real interest rate (Table 5). We estimate in log linear form to get direct point estimates for ε.*(sh)*; as in Table 4, we estimate the perfect foresight (equation (27)) and rational expectations (equation (28)) versions. In both versions, the interest elasticity parameter is not significantly different from zero, even though we will use the point estimates to simulate the interest elasticity of ex-post investments. We have used the log of personal disposable income for the income variable.

**Greece: Credit Availability—Estimates, 1964–86**

*c*refers to the constant term;

*y*denotes personal disposable income;

*r*denotes the deposit rate by banks;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson statistic; and * denotes significance at 95 percent.

**Greece: Credit Availability—Estimates, 1964–86**

Coefficients | |||||
---|---|---|---|---|---|

Long-Term | |||||

Borrowing (log b) | C | log (y) | log (1 + r) | R^{2} | D-W |

Equation (27) | −5.4760 | 2.0690 | 0.4077 | 0.79 | 1.73 |

OLS (t-ratios) | (12.85*) | (5.64*) | (0.20) | ||

Equation (28) | −5.4360 | 2.0342 | 0.1376 | 0.71 | 1.72 |

IV (t-ratios) | (3.75) | (3.67) | (0.02) |

*c*refers to the constant term;

*y*denotes personal disposable income;

*r*denotes the deposit rate by banks;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson statistic; and * denotes significance at 95 percent.

**Greece: Credit Availability—Estimates, 1964–86**

Coefficients | |||||
---|---|---|---|---|---|

Long-Term | |||||

Borrowing (log b) | C | log (y) | log (1 + r) | R^{2} | D-W |

Equation (27) | −5.4760 | 2.0690 | 0.4077 | 0.79 | 1.73 |

OLS (t-ratios) | (12.85*) | (5.64*) | (0.20) | ||

Equation (28) | −5.4360 | 2.0342 | 0.1376 | 0.71 | 1.72 |

IV (t-ratios) | (3.75) | (3.67) | (0.02) |

*c*refers to the constant term;

*y*denotes personal disposable income;

*r*denotes the deposit rate by banks;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson statistic; and * denotes significance at 95 percent.

Table 6 reports results of simulations for ε(*i*), the interest elasticity of total investment, with α = 0.33. This is actually the sample average for bank lending as a proportion of total investment.

**Simulations for ε( i) Estimated Interest Elasticity of Investment**

**Simulations for ε( i) Estimated Interest Elasticity of Investment**

Real | ||
---|---|---|

Interest Rate | (1) | (2) |

−0.10000000 | −0.04561265 | −0.05895918 |

−0.05000000 | −0.05562118 | −0.06475735 |

0.00000000 | −0.06562972 | −0.07055553 |

0.05000000 | −0.07563825 | −0.07635371 |

0.10000000 | −0.08564679 | −0.08215188 |

0.15000000 | −0.09565532 | −0.08795006 |

0.20000000 | −0.10566386 | −0.09374824 |

0.25000000 | −0.11567240 | −0.09954641 |

0.30000000 | −0.12568093 | −0.10534459 |

0.35000000 | −0.13568947 | −0.11114277 |

0.40000000 | −0.14569800 | −0.11694094 |

0.45000000 | −0.15570654 | −0.12273912 |

0.50000000 | −0.16571508 | −0.12853730 |

**Simulations for ε( i) Estimated Interest Elasticity of Investment**

Real | ||
---|---|---|

Interest Rate | (1) | (2) |

−0.10000000 | −0.04561265 | −0.05895918 |

−0.05000000 | −0.05562118 | −0.06475735 |

0.00000000 | −0.06562972 | −0.07055553 |

0.05000000 | −0.07563825 | −0.07635371 |

0.10000000 | −0.08564679 | −0.08215188 |

0.15000000 | −0.09565532 | −0.08795006 |

0.20000000 | −0.10566386 | −0.09374824 |

0.25000000 | −0.11567240 | −0.09954641 |

0.30000000 | −0.12568093 | −0.10534459 |

0.35000000 | −0.13568947 | −0.11114277 |

0.40000000 | −0.14569800 | −0.11694094 |

0.45000000 | −0.15570654 | −0.12273912 |

0.50000000 | −0.16571508 | −0.12853730 |

As is evident from Table 6, the interest elasticity of total investment is negative, even at real interest rates as low as -10 percent, for both sets of point estimates. We had essentially the same results when simulations were done for actual rather than average levels of *r* and of α these are not reported here but yielded negative estimates for the entire period 1964–86.

It may be useful to compare these results with a direct estimate of the interest elasticity of investment in the private manufacturing sector. To do this, we estimate the reduced form in log-linear form; the estimates are reported in Table 7 below. We included, as before, the log of personal disposable income, which affects credit availability; in addition, we included a measure of real wage increases, *d*log(*w*); the coefficient of this turns out to be positive, which possibly captures the “income effect” of real wage increases on investment. In addition, it appears that investment “overreacts” to current inflation; equation (29) demonstrates the fact that inflation has a positive effect on real investment over and above its effect on the real cost of debt servicing.

**The investment Function**

*c*refers to the constant term; rev is post-tax revenues,

*y*denotes personal disposable income; τ is the lending rate;

*w*is the index of manufacturing wages;

*inf*denotes inflation (percent change in consumer price index); MA(1) refers to the first-order moving average, and MA(2) refers to the second-order moving average,

*R*

^{2}= 0.947; the Durbin-Watson statistic = 1.94.

**The investment Function**

Investment | c | log(rev) | log(y) | log(l +r) | dlog(w) | log(1 + inf) | MA(1) | MA(2) |
---|---|---|---|---|---|---|---|---|

Equation (29) | −0.161 | −1.982 | −1.201 | −1.94 | 2.519 | 0.996 | 0.17 | −0.06 |

OLS | (0.27) | (5.112) | (3.246) | (2.689) | (3.921) | (2.201) | (0.55) | (0.18) |

*c*refers to the constant term; rev is post-tax revenues,

*y*denotes personal disposable income; τ is the lending rate;

*w*is the index of manufacturing wages;

*inf*denotes inflation (percent change in consumer price index); MA(1) refers to the first-order moving average, and MA(2) refers to the second-order moving average,

*R*

^{2}= 0.947; the Durbin-Watson statistic = 1.94.

**The investment Function**

Investment | c | log(rev) | log(y) | log(l +r) | dlog(w) | log(1 + inf) | MA(1) | MA(2) |
---|---|---|---|---|---|---|---|---|

Equation (29) | −0.161 | −1.982 | −1.201 | −1.94 | 2.519 | 0.996 | 0.17 | −0.06 |

OLS | (0.27) | (5.112) | (3.246) | (2.689) | (3.921) | (2.201) | (0.55) | (0.18) |

*c*refers to the constant term; rev is post-tax revenues,

*y*denotes personal disposable income; τ is the lending rate;

*w*is the index of manufacturing wages;

*inf*denotes inflation (percent change in consumer price index); MA(1) refers to the first-order moving average, and MA(2) refers to the second-order moving average,

*R*

^{2}= 0.947; the Durbin-Watson statistic = 1.94.

The estimate of the interest elasticity of investment is -1.94, which is considerably higher in absolute value than our simulations in Table 6, We had used the point estimates for the interest elasticity of bank credit, even though this was statistically insignificant; setting this equal to zero yields estimates of ε(*i*) much closer to the estimate here: the average value is -1.654 for the sample period.

For the second empirical version, we estimated the “steady-state” relation between real corporate savings, profitability, and borrowing as shown in Table 8.

**Greece: Corporate Savings—The Long Run, 1963–85**

*c*refers to the constant term,

*rev*denotes post-tax revenue in the mining and manufacturing sectors;

*b*denotes long-term borrowing; MA(1) is the first-order moving average; MA(2) is the second-order moving average;

*R*

^{2}denotes the coefficient of determination; D-w is the Durbin-Watson statistic; and * denotes Significance at 95 percent.

**Greece: Corporate Savings—The Long Run, 1963–85**

Coefficients | |||||||
---|---|---|---|---|---|---|---|

Long-Run Corporate | |||||||

Savings Function | c | rev | b | MA(1) | MA(2) | R^{2} | D-W |

Equation (30) | 0.1457 | 0.1772 | −0.9591 | 0.82 | 0.20 | 0.81 | 1.93 |

OLS (t-ratios) | (1.45) | (2.29*) | (4.86*) | (2.21*) | (0.46) |

*c*refers to the constant term,

*rev*denotes post-tax revenue in the mining and manufacturing sectors;

*b*denotes long-term borrowing; MA(1) is the first-order moving average; MA(2) is the second-order moving average;

*R*

^{2}denotes the coefficient of determination; D-w is the Durbin-Watson statistic; and * denotes Significance at 95 percent.

**Greece: Corporate Savings—The Long Run, 1963–85**

Coefficients | |||||||
---|---|---|---|---|---|---|---|

Long-Run Corporate | |||||||

Savings Function | c | rev | b | MA(1) | MA(2) | R^{2} | D-W |

Equation (30) | 0.1457 | 0.1772 | −0.9591 | 0.82 | 0.20 | 0.81 | 1.93 |

OLS (t-ratios) | (1.45) | (2.29*) | (4.86*) | (2.21*) | (0.46) |

*c*refers to the constant term,

*rev*denotes post-tax revenue in the mining and manufacturing sectors;

*b*denotes long-term borrowing; MA(1) is the first-order moving average; MA(2) is the second-order moving average;

*R*

^{2}denotes the coefficient of determination; D-w is the Durbin-Watson statistic; and * denotes Significance at 95 percent.

This estimation yields a point estimate for *s*_{2} of 0.9591; thus, the effect of an autonomous increase in credit availability on investment is measured as 0.0409. In addition, the size of the estimated coefficient suggests that, on average, the private sector expects a large effective subsidy rate on its borrowing—the point estimate is 24.4, with the approximation of τ = 0,

Table 9 reports a simple dynamic version of the process of adjustment. Suppose *cs**(*t*) = *i**(*t*) - *b*(*t*) is the “steady-state” level of intended investment; and *dcs*(*t*) is defined as the discrepancy between actual and intended levels of corporate savings. We model the current discrepancy as responding to past levels of discrepancies (following from the dynamic error-correction rule) and to current shocks. We restrict the dynamic specification to a first-order autoregression, whereas current values of *q* and (expected) real wages represent shocks in factor prices. We are unable to employ an appropriately richer specification because of the small number of observations.

**Greece: The Adjustment Process for Corporate Savings**

*c*refers to the constant term; dcs(-1) denotes change in corporate savings, lagged one period, w′ is the (forward) index of manufacturing wages; w″ is the (current) index of manufacturing wages; q is the index of capital goods prices,

*r*denotes the lending rate;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson Statistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

**Greece: The Adjustment Process for Corporate Savings**

Coefficients | ||||||||
---|---|---|---|---|---|---|---|---|

Change in | ||||||||

Corporate Savings | c | dcs(-1) | w′ | w″ | q | (1+r) | R^{2} | D-W |

Equation (31) | 0.1583 | 0.3991 | −0.0126 | — | −0.0214 | −0.1125 | 0.53 | 1.87 |

OLS (t-ratios) | (1.88**) | (1.79**) | (1.24) | — | (0.54) | (1.70**) | ||

Equation (32) | 0.1616 | 0.3785 | — | −0.0190 | — | −0.1342 | 0.53 | 1.87 |

IV it-ratios) | (2.16*) | (1.85**) | (2.54*) | (2.01**) |

*c*refers to the constant term; dcs(-1) denotes change in corporate savings, lagged one period, w′ is the (forward) index of manufacturing wages; w″ is the (current) index of manufacturing wages; q is the index of capital goods prices,

*r*denotes the lending rate;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson Statistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

**Greece: The Adjustment Process for Corporate Savings**

Coefficients | ||||||||
---|---|---|---|---|---|---|---|---|

Change in | ||||||||

Corporate Savings | c | dcs(-1) | w′ | w″ | q | (1+r) | R^{2} | D-W |

Equation (31) | 0.1583 | 0.3991 | −0.0126 | — | −0.0214 | −0.1125 | 0.53 | 1.87 |

OLS (t-ratios) | (1.88**) | (1.79**) | (1.24) | — | (0.54) | (1.70**) | ||

Equation (32) | 0.1616 | 0.3785 | — | −0.0190 | — | −0.1342 | 0.53 | 1.87 |

IV it-ratios) | (2.16*) | (1.85**) | (2.54*) | (2.01**) |

*c*refers to the constant term; dcs(-1) denotes change in corporate savings, lagged one period, w′ is the (forward) index of manufacturing wages; w″ is the (current) index of manufacturing wages; q is the index of capital goods prices,

*r*denotes the lending rate;

*R*

^{2}denotes the coefficient of determination; D-W denotes the Durbin-Watson Statistic; * denotes significance at 95 percent; and ** denotes significance at 90 percent.

Equations (31) and (32) underline the relative importance of persistence in deviations from the steady state via the autoregressive term. At this point we have basically 19 points of data, whereas the estimation of “error-correction” models of adjustment to equilibrium needs much larger parameterizations, and a correspondingly large data set. We use this last version as essentially indicative in nature, which can be set up as a framework to measure the longer-run effects of changes in real wages and in real interest rates.

## III. Conclusions

We analyzed the investment decisions of firms subject to a borrowing constraint. It is widely believed that the low and, in particular, negative lending rates maintained by the banking sector in Greece have generated severe constraints on the availability of credit, and this may have accounted for the actual decline in real investment in the manufacturing sector.

If borrowing is indeed the major constraint on investment expansion, we should expect to observe an increase in corporate savings rates and, hence, a decline in the ratio of borrowed funds to total investment. This, in fact, is not what happened. Aggregate data clearly demonstrate the rise in the “debt/ investment” ratio; analysis of cross sectional data (at the two-digit level of classification) by Tsoris (1984) bears out that for the period 1958–81, there was an increase in this ratio, on average. Most of this increase was accounted for by major declines in the retained earnings rates of a few industries, which are precisely the ones that reported secular declines in their profitability rates.

This analysis suggests that interest rate management will be of limited effectiveness in increasing investment. We say “limited,” because, at the aggregate level, the empirical exercise predicts a negative effect, even though a more detailed analysis of industry-specific data would show up the differential effects on individual industries. When firms are credit-constrained, the rate of return on capital, *r**, is in general not equal to the interest rate, and the size of this discrepancy can also vary quite substantially between industries. As a result, the effect of interest rate increases can differ, at the margin, across industry groups, and quite substantially so.

The analysis here entailed that the interest elasticity of investment depend on the interest elasticity of household savings. Our estimates of this parameter are small and statistically insignificant. Giovannini’s (1985) findings were more or less the same with a large cross-country study. One caveat should be kept in mind: it is presumed that the real interest rates in our study were significantly lower than the “world” rate. If the increase in the domestic rate allows it to catch up with the world rate, there should be a (discontinuous) jump in credit availability, at least in theory, because of the inflow of foreign capital; presumably, ε*(sh)* is infinite at this point. Alternately, we may think of this analysis as locally valid for *r** less than the world rate.

In an earlier paper, van Wijnbergen (1983) also suggested that increases in the bank rate need not increase investment in the “medium run.” This argument was based on the supply of loanable funds. If households can lend directly to firms—in other words, if an unorganized money market provides secondary financial intermediation—the total funds available for investment can actually decrease if they go through banks that have to observe reserve requirements. Our argument is based more directly on the ex-ante investment decisions of firms; among other things, the short-run effect will in fact persist, and does not peter out in a “long-run” analysis.

This solution, of course, begs the question of why profitability rates have been decreasing, in spite of a fairly long period of declining investment. The analysis shows that the long-term increase in the product-wage rates must be one of the important factors, by virtue of both its statistical explanatory power in alternative parameterizations as well as its persistence over a fairly long period. The increase is a consequence of directly redistributive policies followed by a succession of governments. A major policy instrument has been expansion in public sector employment at rising real wages.

The interesting questions that this analysis opens up relate to optimal, or effective, financial reform methods that complement redistributive policy. In particular, we know, at least in theory, that in the presence of full equity markets, an increase in the wage bill can generate positive income effects on investment. It is possible that the transition to perfect capital markets can be more effectively achieved with reforms in equity, as opposed to debt markets, when accompanied by redistributive policies.

## Comment

### Mario Arcelli

Since the beginning of my academic career, when I translated Professor Papandreou’s book, *Economics as a Science*, into Italian, my estimation of Greek economists has been very high. I must say that the paper presented by Dutta and Polemarchakis fully lives up to my expectations, being a fine piece of economic and econometric analysis. Therefore, I feel honored to have been invited by the International Monetary Fund to comment on it. At the same time, I feel a bit uneasy about having to talk about Greek fiscal policy, which is hardly mentioned in the paper and can therefore be referred to only indirectly.

Dutta and Polemarchakis state in their paper that lending rates set by the Bank of Greece have been predominantly negative in real terms since 1973. Credit rates have been subsidized, and administrative controls on credit have been imposed. As a result, they say, one would expect credit availability to be restricted and private industrial investment to be constrained by the shortage of funds. However, survey data from Greek firms in the private manufacturing sector indicate the opposite. According to Deleau (1987),^{1} most individual firms do not perceive a shortage of funds.

For the authors, this appears to be a puzzle, since it implies that even negative real interest rates do not provide an effective subsidy. Therefore, in order to solve the apparent puzzle, they provide us with an elaborate model of the process of investment decision making by Greek firms that attempts to reconcile the facts.

One should note before commenting on the model that conditions in the Greek economy are changing quickly, so that the Dutta-Polemarchakis analysis appears to be more relevant to past experience when interest rates were subsidized and there were credit ceilings than to the future. As the Annual Economic Report of the European Economic Community (EEC) for 1987–88 observes:

Substantial progress has been made in reforming the financial system. Preferential interest rates for certain categories of operators have been abolished. The general level of interest rates, which previously had been lower than the rate of inflation, has been raised and commercial banks have been granted a degree of freedom in fixing their lending rates. The strict administrative rules and regulations governing the administration of credit have been made more flexible and efforts have been made to place treasury bills and medium-term paper with private non-banks. Continuation of these efforts, together with the creation of an efficient non-bank financial market, should facilitate the implementation of monetary policy and contribute to the modernization of the financial system, which is essential for the development of the country.

^{2}

So much for the future. Coming back to the Dutta-Polemarchakis paper, one is faced with the problem of explaining why, in spite of negative real rates of interest, firms have not tried to borrow more at these rates and do not appear to have been constrained by financial shortages in their investment choices. The authors argue that bank credit is not the only source of investment finance and that firms use profit retention as the source of residual finance for their investment outlay. Corporate saving reveals itself to be the most important source of investment financing and might explain, according to Dutta and Polemarchakis, the puzzling apparent lack of perceived fund constraints. Corporate saving provides additional funds, which in many cases substitute for external financing; the choice between the two channels of investment financing would be consistent with an optimization process for finding the source of financing. The model of credit-constrained investment financing proposed by the authors is different from a simple model where firms react passively by setting their investment budget equal to the credit ceiling. In the Dutta-Polemarchakis model a firm derives its investment budget, its profit retention rate, and its dividend payout simultaneously, taking into account the credit ceiling and debt-servicing costs. Any firm behaves as a profit maximizer, and a representative shareholder maximizes a utility function in the two variables: dividend income and anticipated profits, which are inversely related to debt-servicing costs. Corporate saving, which is given by real retained profit, adds up to credit to provide investment financing. Increasing the credit limit would go, partly, toward increasing the dividend payout ratios.

In the Greek experience, according to the estimation of the model, the effect of an autonomous increase in credit availability on investment is measured by a coefficient equal to 0.0409. This result can be read as a proof of the relative lack of constraints on investment.

At this point, several possible explanations of the facts could be offered that have not been explored, or at least mentioned by the authors. One explanation could be provided by developments in investment in Greece over the past several years. Growth in gross fixed investment in real terms in plant and equipment was -8.2 percent of gross domestic product (GDP) in 1983, -0.9 percent in 1984, 4.4 percent in 1985, -6.5 percent in 1986, and 2.2 percent in 1987; whereas total investment growth was negative in each of those years except 1985. It is therefore not surprising that firms have not perceived any shortage of funds, even with credit ceilings. The growth path of investment has been slow (negative), because the rates of profitability have been declining as a consequence of strong increases in real wage rates and, in the more recent period, as a consequence of the stabilization policy. Of course, the empirical data can be reconciled with the results of the Dutta-Polemarchakis model, but it is not necessary to resort to such an optimizing process to explain the lack of shortage of funds for investments.

Another possible explanation is that the choice of investment projects might have been restrained by a high degree of uncertainty about the future, which required a compensating premium higher than that granted by a negative real interest rate. In such circumstances, entrepreneurs prefer to finance investment with corporate saving rather than engaging in ambitious projects with borrowed funds, which carry the risk of worsening the firm’s financial structure and profitability. In this case, retained profits are not a residual source of investment financing. It is not so clear, at least to me, from the figures given in the text, that corporate saving is a residual source of financing.

Other possible explanations for the lack of a perceived shortage of investment funds include the absence of competitive market conditions and nonoptimal investment decisions.

It would be interesting to know whether the authors’ choice was dictated by their evident preference for an elegant economic model or by their collection of convincing empirical evidence of generalized optimizing behavior by firms. The assumption of perfect foresight or of rational expectations concerning real wages and real interest in order to analyze the corporate savings function remains questionable—at least, to me.

But let me now return to some of the results of the model within the perspective of fiscal policy. It appears from the estimation of the model that a relaxation of the credit ceiling that was not induced by higher interest rates would lead to some improvement, if modest, in investment.^{3} This result, as I said earlier, can be read as a proof that firms did not suffer from investment financing constraint. In addition, real interest rates were negative. Would Dutta and Polemarchakis then accept the conclusion implicit in the previous statements, that in Greece private investment was not crowded out by public sector expenditure? How to reconcile this conclusion with the fact that for a number of years the public sector in Greece has been expanding rapidly with a significant increase in public debt?

One could argue, on the one hand, that persistent high deficits in the current account of the balance of payments, which ranged from 4.7 percent of GDP in 1983 to 4.1 percent in 1984, 8.2 percent in 1985, 5.4 percent in 1986, and 4.2 percent in 1987, provided an additional source of real resources. Capital inflows from abroad have certainly helped to sustain an unstable path. But, on the other hand, if viewed from a structural perspective, the basic problem facing the Greek economy from 1980 to 1985 appears to have been the continuing decline in productive investment, brought about by high wage rate growth and an unfavorable climate for industrial activity associated with a wide-ranging regulation of markets. The fall in actual and expected profitability can therefore explain the negative rate of growth of productive investment.

The expansion of public expenditure has not been positive in this context, if one embraces the hypothesis that unemployment in Greece was of a classical rather than a Keynesian nature, following Malinvaud’s (1985) classification.^{4} One passage in the paper seems to support this interpretation. Noting the decline in profitability growth rates, the authors say that their empirical analysis “suggests that one of the major determinants of this decline… is the long-term increase in product-wage rates in manufacturing …” (p. 162).

Elements of repressed inflation were also present. The situation was made more complex in Greece by the deficit on the current account of the balance of payments, and by the rate of inflation—after fluctuating around 20 percent for a number of years, consumer prices were growing at a rate of 25 percent at the end of 1985.

Greek economic policy in the last few years, starting in 1985, has been quite appropriate; it is based on the prescriptions of a stabilization plan, supported by the EEC, designed to achieve an improvement in the balance of payments through a deliberate cutback of domestic demand, in order to stabilize the level of the external debt.

Wage policy, the key instrument for restraining domestic demand, was based on a system of degressive indexation focused on a targeted inflation rate and excluding the effects of import prices. In addition, only moderate growth was permitted for farm incomes. The strict incomes policy led to a 10 percent reduction in real wages in the period 1986-87.

Since the end of 1985, according to the EEC’s Annual Economic Report for 1987–88, the government has been working on a phasing-out of price controls. This policy, combined with the moderate growth of wage costs and a consistent exchange rate policy, has contributed to a significant improvement in the financial position of enterprises, which is expected to lead to a significant upturn in productive investment. Thus, in view of the scale of the adjustment of real wage incomes in the past two years, wage policy can no longer be the main instrument for managing demand.

To avoid a deterioration in the balance of payments, reducing the public sector deficit will have to be the prime instrument of demand management. This will be difficult, EEC experts warn, since the social security deficit is tending to increase, and incomes policy will not contribute to the moderation of the public sector wage bill to the same extent as in the two years prior to 1987–88.

I will not discuss further the economic policy suggested for 1988, but it appears evident that the Greek economy is now on a better path than in past years. We must, however, be grateful to Dutta and Polemarchakis for their effort to clarify important issues of investment financing and firm behavior in past years. Their model is also important as an analytical tool.

## Comment

### Christian de Boissieu

This paper deals only incidentally with fiscal policy in a brief reference to the subsidized interest rates prevalent in Greece prior to the recent process of financial liberalization. Its main purpose is to study the behavior of firms facing a quantitative credit constraint and to analyze the role of corporate savings as a means to loosen this constraint and to disconnect the pace of capital accumulation from the constraints concerning credit. Since this paper analyzes the optimizing behavior of firms facing some additional constraint (“additional” in comparison with a general equilibrium solution), it also represents an application of second-best theory to the field of finance and investment.

To summarize some of Dutta and Polemarchakis’s arguments, I find it useful to refer to the following diagram, which features different regimes for firms’ financing structure.

The horizontal rows give information about the situation on the credit market, whereas the vertical columns concern the possibility or the impossibility of self-financing by firms. There are four possible regimes. The core of the argument presented in the paper under review is the following.

In a B-regime, the adjustment is on the saving function, which, in this situation, is the short side of the loanable funds market. Therefore, actual investment is an increasing function of interest rates, since the rise in interest rates induces more saving (in this B-regime, we only have to consider household savings, since there is no room for corporate savings). The B-regime gives the same conclusions as the McKinnon-Shaw model.

In an A-regime, corporate saving has to be considered. In this regime, the short side of the loanable funds market is now the investment function. As a result, actual investment is a decreasing function of interest rates, with the two effects working in the same direction: that is, the traditional Keynesian “profitability effect,” and a “liquidity effect,” by which a rise in interest rates diminishes, other things being equal, firms’ liquidity and the possibility of self-financing.

All this means in Dutta and Polemarchakis’s paper is that self-financing explains the switching from one regime to the other. The discussion presented above concerns only the first row of the diagram. I think it is crucial to answer the following two questions. First, can we assume or confirm the existence of credit rationing? The authors do not insist on this point. Some extra information (I shall come back to this point later) would be useful in the case of Greece as it is for other countries. Second, if there is some credit rationing, where does it come from? It may be the consequence of transitory or permanent credit ceilings, or it may be the consequence of rational behavior by commercial banks. Since credit ceilings have been permanent for a long time in Greece, and since the authors present no reference to the well-known literature on the rationality of credit rationing, I am guessing that they have the first interpretation in mind.

Like most southern European countries (and like France until the end of the 1970s), Greece is an “overdraft economy.” According to Hicks, “In a pure overdraft economy where firms kept no liquid reserves, they would be wholly dependent, for their liquidity, on the banks.”^{1} On more empirical grounds, indirect finance procedures (that is, financial intermediation) are predominant, and capital markets play a residual role. This has been the case in Greece, as suggested by the figures given by the authors, but recent efforts to promote new financial instruments and higher real interest rates will encourage the development of direct finance. For structural reasons extensively developed by work at the Bank of France between 1975 and 1980, overdraft economies incorporate specific features, such as a high degree of rigidity in nominal interest rates; the prevalence of indirect monetary control (credit ceilings); and the quasi-certainty that commercial banks will be refinanced at the central bank (which implies a causation between the money stock and the monetary base that is the opposite of the usual causation implied by money multiplier models). This paper is a useful contribution to the literature on the overdraft economy, which highlights micro-economic aspects, in particular the optimal corporate savings decision (and dividend policy) in an intertemporal model with a permanent credit constraint.

The authors propose for Greece a loanable funds theory of investment, including a hierarchy between the sources of financing, which is close to a lexicographic order. It can be called a loanable funds theory, since, during the period, the amount of investment is determined by the sum of the credit limit and retained profits. There is a lexicographic order for the financing, since retained profits intervene as a means of financing when the credit constraint is binding, and not before. This analysis raises several questions.

Is it realistic to assume that the credit constraint is always binding? Even in an economy with permanent credit ceilings and real returns on capital goods well above (structurally negative) real interest rates, we may observe an excess supply of credits at current interest rate levels (instead of an excess demand) due to such developments as low real growth and low expected demand for goods. Then, as I said earlier, it would be interesting to use extra information to determine whether credit ceilings are binding or not. The French experience with credit ceilings showed that penalty rates on parallel credit markets and short-run movements in the transaction velocity (or a related measure) can sometimes be used as relevant proxies for the state of equilibrium or disequilibrium on the credit market. It would be valuable to have some of this information for Greece.

Positive leverage due to the positive gap between real return on capital and real interest rates may justify the lexicographic order between external financing (from banks) and internal financing. But this argument must be weighted by the opposite Kaleckian argument of “increasing risk.” In the model, internal financing is complementary to external financing because the credit constraint is effective. My feeling is that, within the credit ceiling, firms may trade off between the two types of financing, taking into account the respective costs (including the opportunity cost of self-financing) and the risk structure.

Instead of focusing on the influence of interest rates on investment, it would be more relevant to study the impact of profitability (that is, the difference between the real rate of profit and real interest rates) on investment. In Greece, before the adoption of the stabilization program in October 1985, the profitability of physical capital was low, and therefore “classical” unemployment represented a high proportion of global unemployment. The main objective of the stabilization program was to raise, through wage disindexation, the share of profit in national income and to restore profitability—that is, to raise increases in the real rate of profit above the rise in real interest rates caused by disinflation and financial liberalization.

In the model which is proposed by Dutta and Polemarchakis, commercial banks are rather passive. This assumption seems correct for stylizing the functioning of a financial system that is traditionally “overdetermined” the Greek monetary authorities used to fix interest rates and the volume of credit. Notwithstanding this structural phenomenon, it would perhaps be difficult but enlightening to articulate the macroeconomic constraint on credit (introduced by the monetary authorities), and the microeconomic rationing behavior of commercial banks, mainly based on the assessment of borrowers’ default risk.

In theory and in the empirical estimates, nonlinearities may play an important role, due to the existence of many thresholds. The authors estimate with great ingenuity the effect of an autonomous increase in credit availability, and its split between investment and dividends. The respective weights of the two influences may vary with the level of the capital stock and the relative factor cost (real wages over real interest rates) and the attainment of lower or upper thresholds for these variables. There is also a puzzling gap between the economic analysis and the empirical result. Is it consistent to assume that the credit constraint is binding and restraining capital accumulation, and to show that, according to the data, when it is relaxed, 96 percent of the relaxation is used for dividend payments, and only 4 percent for investment financing?

This paper is a contribution to the analysis of spillover effects from the credit market to the market for capital goods. For a firm facing a credit constraint, the ways to circumvent, at least partially, such a constraint are numerous. For example, firms may resort to external financing. In an overdraft economy, the bond market and the stock market can be used only marginally. Three potential degrees of freedom are available: (1) capital inflows from abroad; the figures show that they represent an increasing proportion of the external financing of Greek private firms (20 percent in 1984); the tightness of exchange controls conditions the use of this channel; (2) trade credit (studies for various Organization for Economic Cooperation and Development (OECD) countries suggest that trade credit is endogenous to the stance of monetary policy); and (3) commercial paper procedures, which may exist even without a formal commercial paper market.

Firms may also resort to internal financing—that is, increases in profitability and corporate savings due to wage disindexation (which is very important in Greece since the adoption of the stabilization program) and increase in prices (for price-making firms).

The authors favor adjustments through profitability and the corporate savings decision. The lesson I draw from the French experience of credit constraints is that we need to develop microeconomic models explaining the choices firms make between alternative ways to circumvent credit constraints. Those models, incorporating traditional constraints and extra constraints concerning the external environment (such as exchange controls), will present second-best solutions adapted to quantitative constraints on certain types of financing. These solutions, dependent on relative costs and interest rate risk, among others, may be a combination of internal and external financing (and of the different channels within each category), rather than one-dimensional. For instance, I suspect that given such a permanent credit constraint as Greece has experienced, informal commercial paper procedures may have developed between firms, in order to finance not only short-term investments but also a part of long-run investments. Since trade credit and commercial paper have a direct impact on the velocity of money, even piecemeal information about the evolution of transaction velocity (or some proxy of this velocity, like the rate of turnover of deposits) could be relevant.

The paper includes puzzling relationships between credit and saving. In the model that leads to empirical estimates, the credit ceiling set by the banking sector is endogenous, and determined by household savings collected by the banking sector. The analysis leaves the world of the overdraft economy and focuses on a system with pure financial intermediation, without net money creation. If the credit ceiling is set by the banking sector on a voluntary basis—that is, without being imposed by the central bank—the model must add to the existing equations other equations representing rational behavior by banking firms, given their assessment of borrowers’ default risk. If the credit ceiling is imposed by the monetary authorities, we are more perplexed: what can be the justification for a credit ceiling policy in an economy where financial intermediaries only transfer private savings, and where there is no net money creation to fill the gap between ex-ante saving and investment? The consistency of the interesting analysis presented by Dutta and Polemarchakis would be, I think, enhanced by a treatment of the relationship between stocks and flows, in particular, between credit and saving, which would be better adapted to the financial system they are studying in this paper.

## Comment

### Augusto Graziani

The purpose of Dutta and Polemarchakis’s paper is to examine the the impact exerted by credit policies (and more specifically by interest rate management) on investment decisions taken by firms in the mining and manufacturing sectors in Greece. The analysis is cast in macroeconomic terms, in that the authors analyze the mining and manufacturing sectors as a whole.

The authors argue that credit policy is important, because, in the case of Greece, in addition to retained profits, bank credit is a major source of investment finance. In fact, they state that finance coming from domestic financial markets has been practically nonexistent. At the same time, they disregard the effects of capital inflows from abroad, which leaves *invested profits* as the only source of long-term finance. On the liquidity side, they neglect altogether the possible presence of money creation due to government deficits. In their model, there is therefore only one source of liquidity—namely, *credit creation by banks*.

The authors assume, both on the theoretical and the empirical level, that private investment is financed by two possible sources only: *long-term bank borrowing and retained profits*, or corporate saving (equation (20c), p. 168). Consequently, the main problem firms are confronted with is to determine the optimal mix of the two possible sources of finance, on the assumption that one of them, namely bank credit, is rationed and that investment decisions themselves depend on the rate of interest.

The authors assume that total household savings are kept as bank deposits. Banks, which are considered as *financial intermediaries*, are assumed to collect deposits and to lend liquidity out to firms, which use it for financing investment.

Empirical investigation leads the authors to the conclusion that investment in Greek industry has been mostly financed by retained profits and only marginally by long-term bank credit. In their view, any increase in bank credit is likely to produce an increase in dividend payments rather than in investment levels. This result, the authors conclude, contradicts the widespread idea that investment activity is severely constrained by credit rationing.

The analysis of corporate profits as a source of investment finance is not objectionable. The authors correctly emphasize the importance of internal finance, and its dependence on the level of interest charges and consequent financial burdens. However, the consideration of bank credit as a second possible source of investment finance is apt to raise more serious questions.

One question arises from the authors’ definition of banks as financial intermediaries, whose principal activities are collecting savings and financing investment. In a theoretical model, there is no problem with defining banks as brokers. This means that the very moment savings are deposited with a bank, liquidity is lost to savers and passed on to firms. In this case, liquidity is actually transmitted from savers to investors, and banks act as real intermediaries.

However, this definition of banking activity becomes illegitimate when it is applied to an analysis of historical data. In present-day practice, bank deposits are in the form of accepted means of payment, and the opening of a deposit implies no loss of liquidity to the depositor. This also implies that any time a bank finances a firm, the bank is not transmitting to it liquidity collected elsewhere, but is actually creating new liquidity without destroying any previously existing liquidity. It is certainly true that a single bank, by collecting deposits, acquires new reserves. But the whole banking system acquires no new reserves just from the movement of deposits from one bank to the next; the only way the system gets extra reserves is to borrow them from the central bank (or acquire them as a consequence of a government deficit). Even if banks could really finance investment (and, as we shall see, this is a very dubious point), they would definitely not finance it by means of liquidity gathered by collecting deposits.

Another implication of the fact that deposits are a means of payment is that as a rule savers do not keep the whole of their savings in bank deposits. (Of course, this can be legitimately assumed. But the implications of a similar assumption should not be forgotten—namely, that if total yearly savings were added to bank deposits, the quantity of money would be increasing each year by the whole amount of households’ savings.) Rather, households tend to consider their bank deposit as a cash balance whose amount is variable, and they will add to it only when its current level deviates from whatever is considered optimum.

Firms borrow from banks in order to finance current production costs. The amount of borrowing therefore depends on production levels and production costs. If we consider firms as a whole, disregarding internal transactions, and assuming a closed economy and no government sector, the amount of borrowing coincides with the wage bill. It has no closer connection with the level of investment than it has with the level of consumption or of government expenditure. Once firms have been financed for the whole amount of their current costs and the wage bill has been paid, two extreme cases can be envisaged.

In the first case, households spend the whole of their money income either on the goods market or on the financial market, where they buy securities issued by firms. In this case, the amount of bank borrowing is always equal to the wage bill. Each period, new liquidity is borrowed from banks, while at the same time an inflow of the same amount comes in from sales of goods or from issues of securities.

In the second possible case, households regularly add a fraction of their income to their bank deposits. In this case, the liquidity that goes back to firms is lower than the liquidity paid out as wages, and the amount firms have to borrow from banks becomes higher and higher over time.

In the first case, bank credit can be defined as short-term credit, not because it is temporary, but because of its nature as a “revolving fund.” In the second case, in each period a fraction of bank credit drops out of the revolving fund, and firms need what might be called permanent or “long-term” bank credit.

In both cases however, the function of bank credit is to finance cash balances held by agents (in the simplified case under examination, where intrafirm transactions are disregarded, total cash balances are held by households), and has nothing to do with investment. The amount of bank credit outstanding is related to the wage bill, to the payments system prevailing in the market, and to households’ demand for cash. (In a pure credit economy, with no central bank and no legal tender, the amount of bank credit will cover the total money stock.) In any case, bank credit is not related to investment decisions or to the requirements of investment financing. Given the wage bill, firms might decide either to run down investments to zero and to produce only consumer goods, without reducing by one penny the amount of their bank debt; or to double investment expenditure by correspondingly reducing the production of consumer goods, without increasing their bank debt.

Although bank debt has no relation to investment, it is closely related to the demand for money. Given the wage bill, any increase in the demand for cash will increase the long-term bank debt of firms, even if investment is constant. In the extreme opposite case in which liquidity preference is zero (all money income of households is immediately spent on goods or securities), firms would have no bank debt, whatever the level of investment. Since bank debt depends on the demand for money and not on investment, a high level of bank debt does not mean that investment is particularly high or that sources of finance other than bank credit are lacking.

Since the function of bank credit is to finance the overall costs of production (or to supply cash balances required by agents, which is the same thing) and not to supply investment finance, investment can only be financed on the financial market or with firms’ profits. If, as in the case of Greece as described by Dutta and Polemarchakis, financial markets are practically nonexistent and if foreign capital is disregarded, retained profits are the only possible source of investment finance. If banks were to finance investment, they could not do so by supplying credit, but by buying capital goods themselves, something they could only do by using profits earned from banking activities. But this is again a special case of investment being financed through corporate profits.

The above considerations can be restated in a somewhat simplified way in terms of standard macroeconomic theory. The investment plans of an individual firm can be financed by bank credit. But at the moment investment is performed (a new plant is built, or new capital goods are produced), liquidity made available by banks becomes income for the recipients of investment expenditure. The increase in income will continue until additional savings equal additional investments. Therefore, in any ex-post situation, investment is always financed by saving no matter what form it takes, whether domestic or foreign (equal to a deficit in the trade balance). In any case, it is not bank credit, no matter how the initial investment expenditure has been financed.

Because investment, like any other kind of expenditure, can (and probably must) be initially financed by bank credit, whereas final financing can only be supplied by saving, no empirical analysis of statistical data will ever find that investment has been financed by bank credit. This is another way of repeating the old Keynesian truth that in any ex-post situation savings and investment are necessarily identical.

It is of course possible, or even probable, that more generous bank credit may make possible a higher level of investment expenditure. Availability of bank credit can therefore remove a constraint, or even be a stimulus to higher investment, which will bring about a higher level of money (or possibly real) income. But once investment is performed, two things will happen: the higher investment will be matched by higher saving; and bank credit will be equal to the existing money balances and, although satisfying the current demand for money, will not be financing investment.

Since the function of bank credit is to supply ex-ante financing to cover overall costs of production, no direct correlation can exist between bank credit and investment. It is perfectly legitimate to assume, as the authors do, that bank credit is spent wholly or partially on investment projects. But since ex-post investment finance is necessarily supplied by savings, no empirical relationship between investment levels and the amount of bank credit can be found in statistical data.

Of course, if the relationship between the money stock and nominal income is stable, and if the investment multiplier is also stable, bank credit will be positively related to investment. But no matter how successful, any similar statistical test will always be measuring the demand for money and not the financing of investment. Any attempt to test the existence of a direct relationship between bank credit and investment is devoid of any theoretical basis.

## References

Deleau, M., “Industrial Investment in Greece: Analysis and Recommendations” (Athens: Ministry of National Economy, 1987).

Giovannini, Alberto, “Saving and the Real Interest Rate in LDCs,”

(Amsterdam), Vol. 18 (August 1985), pp. 197–217.*Journal of Development Economies*International Monetary Fund,

*Theoretical Aspects of the Design of Fund-Supported Adjustment Programs*,*Occasional Paper No. 55*(Washington: International Monetary Fund, 1987).Malinvaud, Edmond,

*Profitability and Unemployment: Based on the Marshall Lectures Given at the University of Cambridge, 1978*(Cambridge: Cambridge University Press, 1980).Malinvaud, Edmond, The Theory of Unemployment Reconsidered (New York: Blackwell, 2nd ed., 1985).

Organization for Economic Cooperation and Development,

*OECD Economic Surveys: Greece*(Paris: OECD, various years).Sargan, J.D., “The Consumer Price Equation in the Post War British Economy: An Exercise in Equation Specification Testing,”

(Edinburgh), Vol. 47 (January 1980), pp. 113–35.*Review of Economic Studies*Tsoris, Nicholas D.,

*The Financing of Greek Manufacture*,*Centre of Planning and Economic Research Studies No. 8*(Athens: Centre of Planning and Economic Research, 1984).van Wijnbergen, S., “Interest Rate Management in LDC’s,”

(Amsterdam), Vol. 12 (September 1983), pp. 433–52.*Journal of Monetary Economics*

^{}1

M. Deleau, “Industrial Investment in Greece: Analysis and Recommendations” (Athens: Ministry of National Economy, 1987).

^{}2

Commission of the European Communities, *Annual Economic Report, 1987–88* (Brussels. October 21, 1987), p, 105.