Demand-pulled innovation under liquidity constraints

Abstract

Using a panel of 211 Italian manufacturing firms for 1995–2001 and the Least Squares Dummy Variable Corrected (LSDVC) estimator recently discussed in the econometric literature, it is shown that demand-pull innovation is particularly significant in liquidity-constrained companies in both the short run and long run.

Acknowledgements

Thanks are due to Capitalia for providing data. This article is part of the research project: ‘Domanda, innovazione e dinamica industriale’ (MIUR prot. n. 2003137229); financial support from MIUR is gratefully acknowledged. Alessandra Catozzella provided excellent research assistance.

Notes

¹ For a more detailed description of the dataset, see Piva and Vivarelli (2006).

² The Arellano–Bond estimator – used for initializing the adopted econometric procedure, see below – on the overall sample of 211 firms does not accept the assumption of endogeneity of sales; the correspondent Sargan test rejects the null hypothesis of validity of instruments under the assumption of endogeneity of sales (χ²(17) = 31.95**).

³ In particular, with an increasing level of accuracy: ; ; .

⁴ In their conclusions, Bun and Kiviet (2003, pp. 151–2) suggest a lower order correction when both N and T are two-digit numbers; however, this is not the case in this article where T is equal to six.

⁵ In our context ‘long-run’ elasticity takes into account the impact of both current and lagged sales, according to the formula . Of course, it would have been better to take lags of higher degree of our main impact variable into account; however, when dealing with short panels, a trade-off exists between the implementation of further lags and the acceptable extension of the time dimension of the dataset used.

⁶ Liquidity-constrained firms are those that replied ‘yes’ to one, two or three of the following questions: ‘Would the firm, at the current interest rate, have asked for more credit?’; ‘In order to have more credit, would the firm have paid a slightly higher interest rate?’ and ‘Has the firm asked for a higher credit line without obtaining it?’

⁷ Four size (<30 employees; between 30 and 50; between 50 and 150; >150) and 21 two-digit sectoral dummies have been included. For computational reasons, it was impossible to insert sectoral and size dummies into the LSDVC estimates; however, their inclusion in the LSDV estimates affect the values and the significance of the relevant coefficients only negligibly. Results concerning the coefficients of time, size and sectoral dummies are not reported.