Financial flexibility, corporate investment and performance: evidence from financial crises

Similar views have also been put forward by earlier studies. For example, the pecking order theory of capital structure, proposed by Myers (1984) and Myers and Majluf (1984), is based on the assumption that firms preserve financial slack to avoid the need for external funds in financing future investment opportunities. Also, Froot et al (1993) suggest that firms maintain financial flexibility to avoid the costs of underinvestment.

A number of studies emphasize the importance of obtaining financial flexibility through low leverage policies (Billet et al. 2007; Byoun 2008; Lins et al. 2010; Campello et al. 2010) or moderate/high cash balances (Opler et al. 1999; Billet and Garfinkel 2004; Almeida et al. 2004; Acharya et al. 2007; Faulkender and Wang 2006; Dittmar and Mahrt-Smith 2007; Kalcheva and Lins 2007; Harford et al. 2008; Riddick and Whited 2008). The main argument of both lines of research is that firms with readily available large cash balances or low leverage can better cope with earnings shortfalls and hence avoid underinvestment.

For example, DeAngelo and DeAngelo (2007) explicitly consider leverage and cash holdings to define financial flexibility and argue that low leverage combined with moderate cash holdings and high dividend payouts constitute an optimal policy regarding flexibility. In line with this view, Gamba and Triantis (2008) show that financial flexibility can be a result of the firm’s strategic decisions regarding its capital structure, liquidity and investment. Moreover, in the light of increased risk in the economic environment, Bates et al. (2008) argue that high cash holdings are related to low levels of debt and hence the simultaneous practice of these policies enable firms to forestall distress and default. Finally, Byoun (2008) reports that small developing firms are more likely to seek financial flexibility and do so through lower leverage and larger cash holdings policies.

Characteristically, between July and November 1997, both Taiwan and Singapore had a current account surplus of more than 10 % and Philippines had almost no deficit although the other countries in the region had a current account deficit of more than 10 % on average. Also, between May 1997 and May 1998 Taiwan was the only East Asian country for which the average daily change in its stock market was not negative (see also Radelet and Sachs 1998; Nixson and Walters 1999).

To identify the pre Asian crisis and Asian crisis period, we follow earlier studies on the subject (see e.g., Lee and Song 2012; Claessens et al. 2006; Lemmon and Lins 2003). For robustness purposes we also adjust the pre crisis, Asian crisis and post crisis to 1995–1997, 1998–1999 and 2000–2007 respectively, for Hong Kong. This helps control for the fact that the crisis occurred in Hong Kong with some delay (Radelet and Sachs 1998; Nixson and Walters 1999; Lam et al. 2010). The results (available upon request) are not affected significantly by this adjustment.

Ownership data cannot be obtained for a small number of firms in our sample. This should not bias, however, our results in a particular manner as there are no statistically significant differences, with respect to their key characteristics, between East Asian firms with and without ownership data at a particular point of time (see also Lins 2003; Lemmon and Lins 2003).

See also La Porta et al. (1998).

In a series of robustness checks that are analytically discussed in Sect. 5, we use industry-adjusted median values for cash and leverage as well as different cut-off points (e.g. the 25th and the 75th percentiles) for classifying firms into different categories. Our results, which are discussed analytically in Sect. 3.4.2, remain robust across the different classifications. .

See Fazzari et al. (1988) and Hubbard et al. (1995) for a discussion on the use of the investment cash flow sensitivity as a proxy for financial constraints.

To decompose the market to book ratio we follow a similar approach to that of Goyal and Yamada’s (2004). Specifically, we regress the MTB ratio against contemporary and lagged sales growth, squared sales growth and industry dummies. The fitted values of this regression are used as a proxy for the fundamental component of stock valuations (MTB ^f), while the residual component is used as a proxy for the residual values (MTB ^r). The inclusion of both components of MTB in the regressions helps capture not only outsiders’ but also insiders’ evaluation of growth opportunities. Alternative ways to avoid the mis-measurement of the proxy of growth opportunities include: (1) the use of contracted capital expenditures alongside MTB in the model (see Carpenter and Guariglia 2008) and (2) the use of an error-correction specification (see Guariglia 2008). Notwithstanding their merit, these methods cannot be utilized in our study due to the nature of our data (e.g. a short-panel) and the lack of availability of information on contracted expenditures. We therefore stick to Goyal and Yamada’s (2004) method to tackle the measurement issue of growth opportunities. Other more straightforward variables that have been suggested as proxies for growth opportunities (e.g. ratio of R&D expenses to total sales, as discussed by Han and Chuang 2011) cannot be implemented in our analysis given the limited data availability (e.g. for the case of Korea, only 10 % of the companies included in our sample disclosed to their R&D expenditures for the year 1998).

The results of the fixed effects model are not qualitatively different from the ones obtained using the random effects estimator. We therefore decide to report only those results that are based on the random-effects estimator (all unreported results are available upon request by the authors).

Recent studies on the inter-temporal nature of financial decisions (see Gatchev et al. 2010) suggest that the lagged value of investment should be included in Eq. (1). However, it is difficult to estimate a well-specified dynamic model using short panels such as ours. Specifically, the requirement to use the lagged values of the dependent and independent variables as instruments makes it difficult to estimate such model separately for all four periods under investigation (pre-crisis, Asian crisis, post Asian crisis and credit crisis) and hence produce directly comparable results. Still, in the spirit of Pindado et al. (2011), we perform a GMM estimation using all firms over the entire sample period (1994–2009). The results show that the adjustment coefficient (given by 1 minus the coefficient of the lagged dependent variable) is above 0.6. One possible explanation for the high value of the adjustment coefficient might be that the costs of deviating from the target are significant, supporting the argument that corporate investment expenditures are persistent over time and firms attempt to sustain their existing policies. We therefore conclude that the substantial differences in terms of investment level and cash flow sensitivity to investment between flexible and inflexible firms in the crisis period, as identified using a specific empirical framework, are less likely to be random and more likely to reflect unexpected changes in the availability of financing.

The weak relation can also be due to the poor empirical performance of q models especially when estimated adjustment costs are excessively large (see, Chirinko 1993 for further discussion on the performance of q-models).

Following Almeida et al. (2004) the KZ index is calculated using the following equation KZ-Index = − 1.002* CashFlow + 0.283* Q + 3.139* Leverage − 39.368* Dividends − 1.315* CashHoldings

Bhaduri (2008) further supports this argument by showing that low dividend payout firms are more likely to be confronted with financial constraints, when compared to their respective counterparts.

The model utilized in Opler et al. (1999) also includes a R&D ratio as an additional explanatory variable. However, we could not obtain reliable information for a large number of firms in our sample and hence do not include R&D in our estimation. Also, we only estimate Eq. (2) for particular years and for each industry separately. This means that there is no need to control for year and industry dummies in Eq. (2).

The utilization of a fixed estimator is not possible given the static nature of some of the variables included in our performance model.

We note that some caution should be taken when interpreting the coefficients on flexibility proxies in Models 3 and 4. We are aware that the positive relationship between the interest coverage ratio and the (leverage) financial flexibility dummy is generated by construction. However, there are reasons why the endogeneity problem should not be as serious as one would suspect initially. First, it should be stressed that the flexibility measure is based on firms’ leverage positions in the pre crisis period and performance is observed during the crisis. Second, as reported earlier, low leverage and hence flexible firms of the pre crisis period increase their leverage substantially in the crisis period. Finally, the positive finding is also in line with the summary statistics regarding the cash flow ratio given that the LL and LL–HC firms have the lowest drops in their cash flows among all the subgroups of firms (see Table 2).

These findings, however, may be mainly driven by the economic conditions that characterized East Asian countries during the crisis period. To this end, an interesting avenue for future research would be the examination of the role of business groups in East Asia within a dynamic setting (i.e. before, during and after the crisis). Indeed, recent studies that focus on Korean firms view the crisis of 1997/1998 as a structural break with respect to the investment behaviour and a shift towards stronger market orientation (see e.g. Rousseau and Kim 2008).

These results hold in models that use different proxies for flexibility (e.g. LL) and/or samples that exclude companies with negative cash flow observations (the results are available upon request).

We thank an anonymous reviewer for suggesting to explore and analyze this alternative interpretation of our findings