ERI Discussion Paper Series No.59
Borrowing Constraints and Role of Land Asset
in Japanese Corporate Investment Decision

April 1994
Kazuo Ogawa
(Kobe University, Economic Planning Agency)
Shin-ichi Kitasaka
(Nagoya City University, Economic Planning Agency)
Hiroshi Yamaoka
(Economic Research Institute, Economic Planning Agency)
Yasuharu Iwata
(Economic Research Institute, Economic Planning Agency)

(Introduction)

One of the salient characteristics of the Japanese economy in the middle of 1980's to early 90's is volatile movement of asset prices, as is typically observed in the land market and the stock market. In the fourth quarter of 1986 the Tokyo Stock Exchange Price Index (TOPIX) stood at 1466.79 and it soared up to 2751.17 in the fourth quarter of 1989. The average rate of increase amounted to 23.3% per annum. As for the land price, the Land Price Index of Six Largest Cities started at 42.6 in September of 1986 and reached 105.1 in September of 1990. The average rate of increase was 25.3% per annum. Soaring asset prices were followed by precipitation of them. The TOPIX fell down to 1276.93 in the third quarter of 1992 at the rate of 26.9% per annum. The land price also fell at the rate of 13.9% per annum from September of 1990 to September of 1993.

In the period of soaring asset prices high growth of GNP was also attained. The growth rate of real GNP exceeded over 4% every year from 1987 to 1991. However, as the asset prices declined continuously, the Japanese economy plunged into deep recession. The growth rate of real GNP was no more than 1.4% in 1992 and it is expected that the growth rate of GNP in 1993 will be close to zero or even negative. Casual observations of the Japanese economy might lead to the assertion that the real economic activities are closely intertwined with the asset prices. This observation has attracted researchers' attention to the role of asset markets in the business fluctuations.

This study is an empirical attempt to examine the relationship of the asset markets with real economic activities in the Japanese economy. In particular we focus on the role of asset markets in business investment decision.

In corporate finance literature Modigliani-Miller theorem is one of the most celebrated theorem according to which the level of investment is independent of the financing method. Underlying the Modigliani-Miller theorem is an assumption of perfect capital market where there is no informational gap among participants in the capital market. However doubt has been cast on this assumption and recent theoretical development along this line demonstrates that when there exists asymmetric information between debtors and creditors then the level of investment is no longer independent of the way of financing and that the firm may face borrowing constrains. 1) The constraints the firm faces, however, can be mitigated by the internal cash flow or/and the collateralizable net worth held by the firm. The upshot is that the level of business investment is no longer independent of the financing method and hinges upon the factors to influence the financial constraints.

It has been frequently asserted in Japan without rigorous analysis that land asset is an important collateral for borrowings. If it is the case, then it is expected that corporate Investment in Japan is much affected by the level of land asset held by the firm. Moreover the surge in corporate investment in the middle of 1980's to early 90's and stagnancy thereafter may be partially attributed to the movements of land prices. Thus in addition to the traditional role of the stock market in corporate investment decision we can find another new channel through which the asset market affects the corporate investment.

We examine whether the firm faces the borrowing constraints in determining the investment level by estimating the first order conditions of the firm derived from the intertemporal maximization problem. The generalized method of moments estimation(GMM) is employed in estimating the first order conditions.2) Our approach has some advantages over the traditional way of estimating reduced-form type investment demand equations. In the typical specification of the reduced-form investment functions, a proxy for future profitability relative to investment cost play a central role among explanatory variables. 3) When the stock market reflects all the relevant information concerning the future profitability of investment, then Tobin's average q, a ratio of the value of the firm evaluated in the financial markets to the replacement value of capital stock, is a sufficient statistics for investment. This approach seems quite appealing since no explicit formulation of future expectations is necessary. However this efficiency assumption of the stock market has been often criticized since the stock price might contain not only the fundamental parts but also the noise uncorrelated with the fundamentals. Then the stock price does not function well in allocating the funds to prospective investment projects in an efficient manner. In fact Ogawa et al.(1993) gives evidence that the stock prices in Japan were contaminated by the noise in the middle to late 1980's and that investment responded differently to the fundamental components and to the noise components.

Marginal q is another candidate to represent the future profitability of investment. Indeed it is free from the implicit assumption of the stock market efficiency underlying the average q, but explicit specification of the future profitability and the discount factor is inevitable. Some kind of arbitrariness always lies in specifying the stochastic process of them and it will affect the constructed series of marginal q.

Direct estimation of the intertemporal first-order conditions by the GMM has no drawbacks stated above and it has the virtue that we can conduct a formal test of the borrowing constraints. Our study is based on the quarterly data reported in the Quarterly Report of Financial Statements of Incorporated Business(abbreviated as QRFS)compiled by the Ministry of Finance. The QRFS is the only data source from which we can construct a quarterly series of land stock. Fourteen industries are chosen as our sample: all industries, construction, manufacturing, food and beverages, textiles, pulp, paper and paper products, chemicals, steel, non-ferrous metal, fabricated metal products, machinery, electrical machinery, wholesale and retail trade, and real estate. The sample period for estimation covers from the second quarter of 1970 to the third quarter of 1990(82 observations).

Our findings are summarized as follows: We could detect the existence of borrowing constraints in half of the industries: all industries, construction, manufacturing, food and beverages, chemicals, steel, and wholesale and retail trade industries. Moreover it is found that the degree of borrowing constraints depends on the asset value of land or the cash flow. Specifically the value of land asset exerts influence on the borrowing constraints in all industries, construction, manufacturing, chemicals, steel, and wholesale and retail trade industries, while the level of cash flow affects the borrowing constraints in all industries, construction, manufacturing, food and beverages, chemicals, and wholesale and retail trade industries. In most of the industries the borrowing constraints were relaxed to a large extent in the period of 1986 to 1990 when the land price rose persistently and the cash flow increased a lot. It should be noted that during this period the growth rate of investment was as high as in the 1960's and early 70's when Japan had attained rapid growth of investment.

The effect of land value or cash flow on mitigating the degree of borrowing constraints is especially large for non-manufacturing sectors such as construction and wholesale and retail trade industries where there are a lot of small-sized firms and hence it is highly likely that asymmetry of information between creditors and debtors might be conspicuous. 4) Our evidence is consistent with the recent development in the investment literature that emphasizes imperfection in the capital market and supports the assertion that land asset has played a collateral role in borrowing.

The paper is organized as follows. In section 2 we present a model of corporate investment decision with borrowing constraints taken into consideration explicitly. Section 3 describes the data set to be utilized, estimation method, and the procedure to test the borrowing constraints. In section 4 estimation results are presented and possible interpretations are provided. Section 5 is a concluding remark.


1)There is a growing body of literature on this issue. For example, see Greenwald and Stiglitz (1988), Gertler and Hubbard (1988), and Bernanke and Gerlter (1989) for theoretical developments. For empirical pieces on this issue, Fazzari, Hubbard, and Petersen(1988), Devereux and Schiantarelli (1990), Hubbard and Kashyap (1992), Whited(1992), and Hubbard, Kashyap, and Whited (1993) utilize the U.S. corporation data and examine the existence of borrowing constrains, while Hoshi, Kashyap, Scharfstein (1990,1991), Asako, Kuninori, Inoue, and Murase (1990), Okazaki and Horiuchi (1992), Miyagawa (1993), and Mitsui and Kawachi (1993) investigate whether the investment spending is bounded by the borrowing constrains for the Japanese firms.

2)There are not so many empirical studies accumulated along this line as those along the reduced-form approach. The studies of Hubbard and Kashyap (1992), Whited (1992), and Hubbard, Kashyap, and Whited (1993) utilize the U.S. corporation data and examine the existence of borrowing constrains by estimating the intertemporal first-order conditions of the firms.

3)Among the studies based on the reduced-form type investment functions, typical are those that use the q variable. However, most of the results are not satisfactory in the sense that there remains sizable residuals unexplained by the q variable and that the residuals exhibit positive serial correlation.
    The variables such as cash flow, sales, and profit variables turn out to be significantly positive when these variables are added to the list of explanatory variables. This might hint the existence of the borrowing constrains. The studies along this line are Blundell, Bond, Devereux (1992), Devereux and Schiantarelli (1990), Fazzari Hubbard, and Petersen (1988), and Hoshi, Kashyap, and Scharfstein (1991). On the other hand, Abel and Blanchard(1986) and Ueda and Yoshikawa (1986) show that persistent residuals are consistent with the q theory when the delivery lags are properly taken into consideration.
    Another approach to reconcile the q theory with the data is to disaggregate the capital stock into several homogeneous parts. This "multiple q" model is formulated theoretically by Wildasin (1984) and Chirinko (1986) and applied to the Japanese corporate data by Asako et al. (1989), and Hayashi and Inoue (1991).

4)Our findings are consistent with the studies of Miyagawa(1993), Mitsui and Kawachi(1993), and Kuroki(1993). They give evidence that some proxy of liquidity is a significant explanatory variable in the investment demand equations for the relatively small-sized firms.


Structure of the whole text

  1. full text別ウィンドウで開きます。(PDF-Format 807 KB)
  2. page1
    1. Introduction
  3. page7
    2. Econometric Issues in Estimation and Testing
    1. page7
      A. Estimation
    2. page8
      B. Testing Procedures
    3. page8
      C. Data Description
  4. page10
    3. Estimation Results and Interpretations
    1. page10
      A. Estimation results of the Euler equations and test of the borrowing constraints
    2. page11
      B. Movement of the borrowing constraints in the phase of business cycles
    3. page14
      C. Related Issues
  5. page15
    4. Concluding Remarks
  6. page17
    Footnotes
  7. page19
    References
  • 1-6-1 Nagata-cho, Chiyoda-ku, Tokyo 100-8914, Japan.
    Tel: +81-3-5253-2111