AccessMyLibrary provides FREE access to millions of articles from top publications available through your library.
THIS paper argues that a serious weakness of empirical work on inventories, at least for manufacturing firms, is that it fails to recognise that inventory investment is but one part of a larger decision nexus addressed by the firm including physical investment, long- and short-term financing, and trade debt and credit. A model is developed in which this wider nexus is taken into account: specifically, the inventory decisions are set in the context of a model where the balance sheet of the firm is modelled by analogy with a portfolio model of assets and liabilities. The empirical application is to two samples of UK quoted companies in 13 different industrial sectors. The focus on firm data is a departure from the existing literature on investment in inventories, which has almost exclusively focussed on aggregate data (see, for example, the literature reviewed in Blinder and Maccini, 1991). The advantage of firm panel data is that the information set is much richer than with aggregate data: the disadvantage is that we have to work with annual data (the only data that are available), and this has consequences for the kind of models that can be tested.
In previous work, we have modelled the behaviour of firms with respect to the main items in the balance sheets by analogy with portfolios of assets and liabilities (Hay and Louri 1989, 1991a, 1991b). In so doing we have borrowed models developed with financial institutions in mind (see Courakis, 1988, 1989). For empirical applications these models require that the utility function of the decision maker be specified, and that the sequence of decision making be identified. We assume that the utility function is a negative exponential in the return on the assets of the firm. This utility function has the usual 'shape', with the implication that the decision makers are risk averse. Given that returns and costs of the various assets and liabilities are stochastic, in determining the optimal mix of assets and liabilities the firm will take into account the variances and covariances as well as the expected returns or costs. In a quoted firm, the decision-makers are the managers of the firm. The shareholders will probably hold diversified share portfolios, and therefore are assumed to be risk-neutral in respect of the operations of the firm. However, the managers will have considerable human capital tied up in the firm, and are therefore likely to be risk averse. Principal-agent theory suggests that (second-best) optimal contracts between the owners of the firm and the managers will not offer the manager complete insurance, and that risk aversion will continue to affect the decisions they make.
It will also be assumed that there is a hierarchy of decisions by the firm, relating to the time horizons over which decisions are made and implemented. Traditionally, investment in physical capital, and raising long term finance (whether equity or debt) have been seen as long term decisions. Inventory investment has been bracketed with bank borrowing, trade debt and trade credit, and working capital, as short term decisions. While this is true in respect of the time horizon of decisions, it should be noted that while inventories and work-in-progress are largely related to production decisions, trade debt and trade credit are more closely related to marketing. The links between these may therefore be quite weak, though both will have an important impact on short-term finance and cash balances.
One precedent for the analysis presented in this paper is the work of Maccini and Rossana (1984). In modelling aggregate investment in inventories of finished products they introduce other stocks as explanatory variables: unfilled orders, work-in-process, raw materials and supplies, numbers of production workers, for each of which items they had aggregate data. Our analysis differs from this in two respects. First, it focusses on interactions with a wider range of stock variables, including financial variables, whereas their analysis is restricted to production. Second, they present no explicit model of the firm behaviour, and prefer to rationalise the estimating equation in intuitive terms; the analysis to be presented in Section 2 is based on a specific model of interactions between firm decisions on balance sheet items. Before turning to that model, we need to review the explanations that have been developed for inventory investment, as a guide to the kind of variables that we should incorporate in our empirical model.
In a review of research on inventories, Blinder and Maccini (1991) drew attention to the empirical inadequacies of the traditional production smoothing/buffer stock model of inventory behaviour, and argued for more empirical attention to be given to the (S, s) model developed by Blinder (1981). The production smoothing/buffer stock model can be simply summarised. Traditional theory of the firm posits rising marginal costs of production. If, therefore, sales vary over time the firm can minimise costs by equating marginal costs in different time periods, subject to the costs of storage. If sales are also stochastic, then inventories serve as a buffer stock. Following Blinder and Maccini (1991), sales fluctuations can be either anticipated by the firm or unanticipated. Three implications then follow: (i) anticipated increases in demand will be met by a mixture of increased sales, increased output and running down stocks; (ii) unanticipated increases in sales will be met only from inventories; (iii) excess inventories will lead the firm to reduce output, increase sales, and carry over some inventories to the next period. Blinder and Maccini point out that (i) implies that output varies less than sales, and (i) and (ii) imply that sales and inventory investment are inversely correlated. Neither prediction is consistent with the evidence for the US. The theory can be rescued by assuming that some shocks arise on the cost side …