By Michael Mahony
Why has there been weak firm-level investment in the UK since the Great Financial Crisis of 2007-2008? Do existing empirical models of firm-level investment provide an adequate explanation for the low levels of investment? These are the key questions which our new ESCoE Discussion Paper ‘Investment and Capacity Utilisation in a Putty-Clay Framework’ examine. To answer these questions, we use a unique dataset, which combines direct measures of firm-level capacity utilisation with company accounts data such as firm-level investment, sales and cash-flow. The direct measure of capacity utilisation is provided by the long-running and well-established Confederation of British Industry (CBI) Industrial Trends Survey (ITS) – which provides firm insights across a wide range of variables.
In standard models of firm-level investment, it is often assumed that firms can freely adjust their stock of capital (or machines). In other words, firms can instantaneously purchase, install, and begin using new machines, as well as reorient existing machines to accomplish new tasks. In reality, this is unlikely. It takes time to purchase, receive delivery and install this new capital – not to mention training employees in the use of new equipment. In addition, existing machines are usually task specific and cannot easily be reoriented to new tasks. A further implication of firms being able to freely adjust their stock of capital is that they never have under-utilised machines – as these will be reoriented to new tasks, reducing variable costs. This is not borne out by the survey data (from the ITS), which shows that 59% of survey responses state firms have under-utilised capital. Assuming firms are free to adjust their capital in the short-run when they are not, overestimates the speed with which capital adjustment occurs. This situation provides an explanation for the weak firm-level investment in the UK since the Great Financial Crisis.
A key contribution of our paper is the estimation of a firm-level investment equation where the capital of firms is fixed in the short-term. Rather than adjusting their stock of machines, firms can instead adjust the utilisation rate of their existing machines (i.e. they change their rate of capacity utilisation). In our paper, we accomplish this by updating the well-known Abel (1981) paper. Abel outlined a theoretical model to explain the relationship between firm-level investment and capacity utilisation. Our paper updates this framework and applies it to our unique dataset. In practice, this involves including capacity utilisation directly in the firm’s production function (by interacting it with the labour stock). This reflects the fact that increased capacity utilisation requires a corresponding increase in employee hours worked (for example through overtime, zero-hour contracts or agency work). If machines are to be used more intensively, then it follows that more employee hours are required.
This framework yields an updated version of the standard model of firm-level investment. Similar to these standard models of firm-level investment, in the long-term, capital is still proportional to sales. This is the capital error correction process, and it captures the breakdown in the long-run relationship between capital and its long-term value (sales). In contrast to the standard models of firm-level investment, our updated version includes a second error correction process – reflecting the relationship between the rate of capacity utilisation and its long-run equilibrium value (the user cost of labour). In the short-term, where firms are unable to adjust their machines, firms increase their rate of capacity utilisation (i.e. they intensify their use of machines) in response to an increase in demand. However, this is only a short-term solution. Increasing the intensity of use wears the machines quicker, and overtime rates for employees (or agency work) are expensive. In the long-term, firms will adjust their capital stock by investing more – therefore meeting demand in a more sustainable way. Thus, our new updated model of firm-level investment accounts for both capital adjusting in the long-term, as well as recognising that firms use their capacity utilisation as a short-term buffer.
In our paper we estimate both the standard model of firm-level investment as well as our updated version. The results are in line with theoretical predictions. However, we show that the adjustment of capital to its long-term value relative to sales is slower in our updated model (where capital is not freely adjustable) compared to the standard models of firm-level investment. In other words, not taking into account the short-term behaviour of firms (where they use the rate of capacity utilisation as a buffer since capital is fixed) overestimates the speed with which capital returns to its long-run equilibrium value. These results are robust to the inclusion of a series of investment constraints such as uncertainty, insufficient finance and poor proposed return on investment.
These results provide an explanation for the sluggish investment we have seen since the financial crisis and why existing models cannot replicate the data. Furthermore, we show that following a simultaneous shock to both capital and capacity utilisation, there is a distinct lack of capital investment. Moreover, if there is a (permanent) negative exogenous sales shock (such as that which was observed in 2009) then capital steadily falls to a permanently lower level in the long-term. This occurs in both the standard model of firm-level investment and in our updated version. However, the speed of this adjustment differs – the standard model of investment adjustment is quicker than our updated model. Thus, in the immediate aftermath of the financial crisis, capital began a long and slow adjustment towards a permanently lower level. This adjustment is even slower if (in the more realistic scenario) firms face fixed factors of production in the short-term and instead adjust their rate of capacity utilisation.
Read the full ESCoE Discussion Paper here.
Michael Mahony is a PhD student at the University of Nottingham and an ESCoE Research Associate.
ESCoE blogs are published to further debate. Any views expressed are solely those of the author(s) and so cannot be taken to represent those of the ESCoE, its partner institutions or the Office for National Statistics.