Euler equation approach Directory UMM :Data Elmu:jurnal:M:Multinational Financial Management:Vol10.Issue2.2000:

working capital on their investment decisions is only significant for the period 1983 – 1985, before the liberalization of the French economy. This result is consis- tent with the hypothesis that the same firms may have faced stronger problems of information asymmetries in the years preceding the expansion of the Paris Bourse.

5. Euler equation approach

5 . 1 . Sample construction The goal of this section is to test the role of borrowing constraints on investment behavior with a structural form that does not include stock market valuation. Results based on the Tobin’s Q approach may be biased, especially if firms do not have access to the same stock exchange markets. For example, companies with bank ties are often linked to the government. For these firms, stock market valuation may not be an important component of their investment decisions. Also, smaller and younger companies may not have a well-established trading market. Consequently, the sample used to test the Euler equation model is slightly different from the one used in Section 4. First, since stock prices are not needed for this approach, a longer time period from 1983 to 1989 can be included in this analysis. Firms are then divided in three sub-samples, using the same criteria as in Section 4.1. The first category consists of 34 groups having strong bank ties. The second category includes 70 groups that do not have bank links. The last sub-sam- ple contains 18 independent firms without significant bank ties. Table 12 compares the financial characteristics of these three classes of firms. As shown in the previous Table 11 GMM estimation of the Q model, 1983–85 versus 1988–90 a Q I,t C WC i,t−1 K i,t−2 Firm category CF i,t K i,t−1 S i,t−1 K i,t−2 Firms without bank ties 0.019 0.092 Coefficient 83–85 0.287 −0.007 0.033 0.028 0.032 0.037 s.e. 0.013 0.096 x 7 2 = 7.940 Coefficient 88–90 0.013 − 0.264 − 0.068 0.487 s.e. 0.019 0.142 0.012 0.059 x 7 2 = 10.280 0.106 0.016 Other markets Model Coefficient 83–85 s.e. Rejected x 7 2 = 15.885 Model Coefficient 88–90 s.e. Rejected x 7 2 = 16.985 a and indicate significance at the 5 and 1 levels. Table 12 Summary statistics by firm categories, Euler equation 1983–1989 a Groups without Independent firms, Groups with bank Variable 126 obs. ties, 238 obs. bank ties, 490 obs. 0.32 0.32 0.34 InvestmentLagK 0.57 0.50 0.57 Cash flowLagK 7.72 7.79 OutputLagK 7.81 Interest coverage 0.19 0.22 0.22 0.08 0.10 Payout ratio 0.08 1.40 DebtLagK 1.47 1.20 1.34 0.84 LiquidityLagK 0.60 0.80 0.45 1.03 Ret. earningsLagK 1.98 1.52 Working cap.LagK 1.21 a Cash flow, operating income+depreciation; payout ratio, dividendoperating income; interest cover- age, interestinterest+cash flow; liquidity = cash+ST securities+accounts receivable−accounts payable; working capital, liquidity+inventory; all ratios reported in this table are median values. Table 13 Euler equation model, 1984–1989 a Groups without Full sample Groups with Independent firms bank ties bank ties 1.355 0.014 1.725 Adjustment cost 2.430 1.198 1.739 2.110 0.523 Parameter a 1.081 1.087 1.110 1.132 Markup 0.058 0.100 0.111 0.043 Parameter m 3.192 1.189 2.914 9.826 x 4 2 = 0.572 0.043 0.880 0.230 Upper tail area a Indicates significance at the 1 level; standard errors are in parentheses; firm specific effects are removed by first-differencing the Euler equation; year dummies are included as regressors and instru- ments in all equations. sections, stocks of liquidity, retained earnings and working capital are much larger for firms without bank ties, especially small and independent companies. 5 . 2 . Model without a debt limit The model used in this section is based on the Euler equation, but it does not include any constraint on the use of external finance. This standard neoclassical model is described in Hubbard et al. 1994. Empirical estimation is made using the Generalized Method of Moments. The estimating equation has to be first-differ- enced in order to remove firm-level fixed effects. Twice-lagged instruments are used in order to ensure that they are orthogonal to the moving average error term. Regression results for this baseline Euler equation estimation are given in Table 13. The neoclassical model without a credit limit is accepted for all group firms, whether they have bank ties or not. However, it is rejected for the category of small and independent companies. This result shows that the hypothesis of perfect capital markets does not hold for firms that are a priori more likely to face information problems. Capital market frictions create a wedge between the cost of internal and external finance, thereby forcing these companies to rely on their cash flow and liquid assets to finance their investment projects. 5 . 3 . Euler equation with borrowing constraints In this section, financial factors are added to the standard neoclassical model by including a limit on the use of debt. The Lagrange multiplier associated with this constraint is then parameterized as a function of the stock of liquid assets. This specification supports the hypothesis that small and independent firms rely more on internal funds to finance their investment projects. Therefore, one would expect increases in the stock of liquid assets to be a sign of stronger financing constraints. The model is described in Appendix A of this paper. Regression results are presented in Table 14. First, the model with a debt limit is rejected for the category of group firms that do not have any bank ties. This finding is consistent with the fact that the baseline Euler equation without a debt limit was accepted for this category. It also validates the evidence from the Q model presented in Section 4 suggesting that these companies are financially uncon- strained. Second, the augmented model is accepted for the category of independent firms. Moreover, the sign of the two parameters linked to the debt constraint multiplier is positive. This result verifies the hypothesis that firms a-priori more likely to face adverse selection problems choose to accumulate liquid assets in response to their poor access to capital markets. Third, the model is also accepted Table 14 Neoclassical model with borrowing constraint, 1984–1989 a Groups without Independent Groups with Full sample firms bank ties bank ties 1.196 1.478 Adjustment cost 2.991 2.536 0.708 1.153 Parameter a 1.870 2.637 1.104 − 11.298 1.221 Markup 1.095 12496.2 0.037 Parameter m 0.148 0.224 0.614 0.240 Parameter a1 1.000 0.830 0.478 0.146 0.434 0.341 − 0.576 − 0.500 Parameter a2 − 0.003 0.252 0.705 1.603 0.270 0.543 − 0.115 − 0.068 Parameter a3 0.002 1.061 0.154 0.362 0.151 0.619 0.426 2.270 x 3 2 = 0.226 49.642 0.518 0.935 Upper tail area 0.000 0.973 a Indicates significance at the 1 level; standard errors are in parentheses; firm specific effects are removed by first-differencing the Euler equation; year dummies are included as regressors and instru- ments in all equations. for group firms with bank ties. However, the sign of the same two parameters is negative for this category of companies. This would indicate that an increase in the stock of liquid assets relaxes the constraint on external finance. Such result is consistent with the literature associating higher debt levels to poor financial health, but inconsistent with the fact that the same firms also accept the model without a credit limit.

6. The role of liquid assets as a measure of asymmetric information problems