There is little to like about recessions – they are periods of substantial dislocation in aggregate economic activity with output falling and unemployment rising. Most firms experience large negative shocks to the demand for their goods and services during recessions, and firm failure rates rise dramatically.
The recent economic history of Japan is no exception. From 1992-2006, Japan suffered through three distinct economic downturns. Even during non-recession years, economic activity did not grow quickly, and the period is often referred to as Japan’s lost decade. 154,000 firms (38%) left the Japanese manufacturing sector and manufacturing employment declined by four million (24.8%).
However, recent work has found that episodes of falling demand are precisely those periods when certain firms increase their innovative activity. Bloom et al. (2015) look at responses to large falls in firm-level demand driven by China’s expansion into Europe and find that surviving firms with the largest exposure to low-wage imports increase, rather than decrease, their innovation activities. They explain this apparent puzzle by developing a ‘trapped factors’ model of the firm where it would rather redeploy resources, in particular labour, away from production and toward research and development (R&D) and innovation (see Bloom et al. 2013, 2014). Workers and other factors are ‘trapped’ inside the firm and, since they are not needed for production, they can be used to create the next generation of products and technology.
Recent work has started to open the black box of the firm and examine resource reallocation within firms. Bernard et a.l (2010) document that most manufacturing output in the US is done at firms that produce multiple products and, in fact, at those that produce products in different aggregate sectors. Their work also shows that surviving firms adjust their product mix frequently over five-year intervals by both adding and dropping products. Kawakami and Miyagawa (2010) examine product switching in Japanese firms over multi-year intervals and find similar results. This focus on the product mix of firms has made the biggest inroads into the large literature on firms and international trade. Aggregate exports are higher in some destination markets because more firms export and exporters ship more products to the foreign markets (see Bernard et al. 2011 and Goldberg et al. 2009).
Our work analyses Japanese manufacturing firms in light of insights from these two recent streams of research (Bernard and Okubo 2015). We ask a simple question: How do firms adjust their product mix over the business cycle? The answer is surprising – recession troughs are the times when it is most likely for continuing manufacturing firms to add and drop products. The fact that firms drop products during recessions seems obvious. Demand is down so weaker products are culled from the mix. However, it is exactly at the same moment of low overall demand that firms are most likely to add products as well.
The evolution of firms and products in Japanese manufacturing
Our work explores the importance of product adding and dropping within manufacturing firms over the business cycle. While recessions have long been associated with declines in output and increased firm exit, we show that they are times of substantial changes in the output mix for continuing firms. Product switching is strongly countercyclical with add and drop rates increasing by more than one-third in recessions. In addition, firms are most likely to both add and drop products around recession troughs.
Our paper documents the mix and evolution of firms and products in the Japanese manufacturing sector from 1992-2006 (Bernaud and Okubo 2015). The study provides evidence on the product mix at Japanese manufacturing firms and the role of product switching in the evolution of aggregate output. The period covered by the firm-product data includes three substantial recessions, a decline in manufacturing as a share of GDP, and an equally sharp reduction in manufacturing employment as a share of total employment.
Every year, Japanese manufacturing firms engage in a substantial amount of product switching – that is, adding and/or dropping products from their output mix. Twenty percent of all firms change their product mix each year by adding and/or dropping at least one product. For multi-product firms that account for more than three-quarters of manufacturing output, product mix changes are even more prevalent and occur primarily by churning; that is, simultaneously adding and dropping products. During transitions from recession to expansion, firm-level product churning increases by 25%. Changes in industry and sector mix show comparable increases during recession transitions.
Figure 1. Product add and drop rates over the business cycle
This increase in product mix changes around recessions is not systematically related to the rise and decline of particular products. Firms do not seem to be leaving some product markets and entering others; 77% of products have higher drop rates and higher add rates during recessions. Furthermore, products with the greatest sunk costs are more likely to be both added and dropped in recessions. For the average product, recessions are times of reallocation across producers. Continuing firms that introduce the product account for a 50% larger share of product output while previous producers and new firms have smaller output shares. Within firms, added products are relatively more important during recessions, contributing 50% more to the firm's overall production.
The contribution of exiting and entering firms to aggregate output changes is small compared to the contribution of intra-firm product adding and dropping in all years. This is especially true during recession transitions when the contributions of product additions by surviving firms are three times as large as those of new firms, while product drops by continuers are more than 2.5 times larger than those of exiting firms.
To explain the cyclical nature of product changes, we appeal to recent theoretical work on ‘trapped factors’ by Bloom et al. (2013, 2014). In this theory, large, negative demand shocks reduce output and leave some production workers underemployed. Since these workers cannot easily be released from employment, the firm uses them to engage in innovation activities. For an economy such as Japan's with a history of lifetime employment, recessions are a time when many firms may face lower opportunity costs of new product development. This leads to increased product dropping (negative demand shocks) and increased product adding (increased innovation) during recessions. We explore the implications of the ‘trapped factors’ model in our data and find some tentative support.
Bernard, A B and T Okubo (2015) “Product switching and the business cycle,” RIETI Discussion Paper 15-E-103.
Bernard, A B, S J Redding and P K Schott (2010) “Multiple-product firms and product switching”, American Economic Review, 100(1): 70-97.
Bernard, A B, S J Redding and P K Schott (2011) “Multiproduct firms and trade liberalization”, Quarterly Journal of Economics, 126(3): 1271-1318.
Bloom, N, M Draca and J van Reenen (2015) “Trade induced technical change: the impact of Chinese imports on innovation, diffusion and productivity”, Review of Economic Studies, February.
Bloom, N, P M Romer, S J Terry and J van Reenen (2013) “A trapped-factors model of innovation”, American Economic Review Papers and Proceedings, May.
Bloom, N, P M Romer, S J Terry and J van Reenen (2014) “Trapped factors and China’s impact on global growth”, NBER Working Paper 19951.
Goldberg, P K, A K Khandelwal, N Pavcnik and P Topalova (2009) “Multiproduct firms and product turnover in the developing world: Evidence from India”, The Review of Economics and Statistics.
Kawakami, A and T Miyagawa (2010) “Product switching and firm performance in Japan”, RIETI Discussion Paper 10-E-043.