Fixed Adjustment Costs and Aggregate Fluctuations


A well-known fact is that individual firms only tend to adjust their employment level, prices or capital stock intermittently, so that these variables follow a pattern of periods of inaction that are punctured by bursts of adjustment. A common model assumption used by economists to explain this ‘lumpy’ firm behaviour is that adjustment comes with a fixed cost; a cost that does not depend on the size of the change, but must be paid for any level of adjustment, however small. This paper analyses how models that incorporate this realistic friction at the microeconomic level predict movements of aggregate macroeconomic variables. The main finding is that aggregate variables – such as the economy-wide employment level, capital stock or price level – move similarly in a model with fixed adjustment costs as in one without, as long as the size of the fixed cost is reasonably small. Hence, in this class of models, despite individual firm behaviour being lumpy, there are no great consequences for economy-wide aggregate variables. This finding is important for several reasons. For example, it is often suggested that fixed costs and intermittent adjustments can generate aggregate price stickiness, which may help monetary policy stimulate demand in the economy. However, the results of this paper casts doubts on this interpretation. 

To illustrate how the main result is reached, the paper sets up a model where firms choose their employment level subject to fixed adjustment costs. The model incorporates firm specific productivity shocks, so that the individual firms occasionally want to adjust their size. The paper shows that firms in this model will follow a so-called Ss-policy: if shocks are small enough firms remain at previous employment levels, but if they are large enough they cause firms to adjust their employment levels. Thus, firms in these models adhere to a lumpy hiring and firing policy, which resembles the one observed in empirical data where more than half of firms do not adjust their employment level quarter to quarter. 

The main challenge in reaching the main result of the paper – that a large presence of inaction at the individual firm level only has a small effect on aggregate dynamics – is to show how the lumpy employment changes of individual firms aggregate when studying all firms simultaneously. The novel idea of the paper is to study the movements in aggregate employment through a related object: the firm size distribution. Since the firm size distribution tells us how many firms operate at any level of employment, aggregate employment can easily be derived from this distribution.  The firm size distribution is also a natural object to use to study the effect of fixed costs and lumpy investments; when firms only adjust their employment intermittently this impedes flows in the firm size distribution as fewer firms leave a point in the distribution at any given moment. However, the paper shows that because of the adjustment frictions, fewer firms also enter any given point in the distribution. It is proven that these forces approximately offset each other, so that movements in aggregate employment are hardly affected by the adjustment frictions, despite the frictions being large enough to cause a substantial amount of firms to refrain from adjusting.  

This approximate neutrality of macroeconomic dynamics with respect to microeconomic frictions has been shown in previous research, however, this paper makes an important contribution by showing that this result holds generally – previously the result has often been reached only for particular numerical model approximations, or for models with a less general form of adjustment frictions. The paper also shows that the result does not depend on equilibrating price adjustments, which earlier research has suggested as a potential explanation. Thus, the paper is able to synthesize and generalize results from the earlier literature. 

This summary was written by Paul Telemo