In an early post on Development Impact, Markus Goldstein noted that one of the notable things about testing surveys was the existential questions that arose, such as “when is a chair a chair?”. This came to mind when thinking about several of the most interesting recent pieces of research I’ve seen on small firms in developing countries, which all get at the question “what makes a firm a firm?”. More generally, at the fluidity of firm boundaries. Of course questions of when activities should take place inside a firm versus in the market go back to at least Coase’s 1937 work (summarized nicely in this Economist piece). But what is new about this recent work is the emphasis on examining empirically how market frictions, risk and volatility, and other features of developing markets affect how firms use capital, labor, and managerial inputs, and may make it hard to say what the size of a firm is. I’ll highlight a few papers as examples:
Capital is fluid across firms
Bassi et al. (2022, ungated) look at carpentry, metal fabrication, and grain milling firms in urban Uganda. They show there is an active rental market for capital equipment, allowing firms operating at a relatively low scale of production to still use large and expensive machinery. The average firm in their sample has 4.5-6 workers, and only 5% have more than 10 employees. But if one were to redefine a firm as all the workers using the same machines, 33% would have more than 10 employees. They note that this fluidity of capital through rental markets therefore allows scale to be achieved collectively in production, even if it is not achieved in each small unit. However, there are transaction costs from renting capital rather than owning it, which they estimate to be about 40% of the direct rental price.
Capital is fluid between a firm and its workers
A new working paper by Morgan Hardy, Jamie McCasland and Jiayue Zhang conducts an experiment with 356 small firms in urban Ghana that are in garment-making and cosmetology (and a few in carpentry). In their baseline survey they find that in these industries it is common for workers to supply some capital to the firm: e.g. garment workers may own the sewing machines they use, beauticians may own their rollers and scissor sets, carpenters may own some of their own tools. They then randomize firms into a control group, a group where the firm owner gets a cash transfer, and a group where one worker who is expected to stay with the firm for at least the next six months gets a cash transfer. The cash transfer is 700 Gh ($254 PPP), about 15% of mean assets, 140% of mean monthly profits, or 7 months of mean worker wages.
They find that when workers receive a cash transfer, over half of them report spending it to purchase some trade-specific capital assets that they then use in the firm that was employing them. They then find that profits increase by an equal amount (about 13%) in the firms where the owner got the cash transfer as in the firms where it was the worker given the cash transfer. Most of the benefits seem to accrue to the firm owner in terms of higher profits, although workers seem to earn a little more when they get the transfer (although it seems one cannot reject equality of wage impacts for the two treatments either).
An interesting implication here is that when workers leave the firm, the firm not only loses labor, but also loses the capital stock that belonged to the worker. So some of the physical capital is tied to workers. This reminded me of my work on long-term impacts of improving management in India, where we found some of the management practices appear to be tied to specific managers, so that when these managers leave, the company stops using the new management practices.
How fluid is labor?
My sense is that we still don’t have great information on the extent to which small firms share the same pool of workers – e.g. with workers working some days or some weeks in one firm, and another day or week in another in the same industry. High frequency matched worker-firm data would be needed to really look at this, which is not available for informal firms and informal workers. In medium-sized Colombian firms in the autoparts sector, we did look at these worker movements to see whether there were spillovers from management improvements through worker movements. Over a 5-year period with monthly employment data linked to firms, we had 22,884 workers who only worked at some point for one of the 159 firms in our sample, and only 272 workers (1%) who worked for two or more of the firms. So at least in that context labor was not very fluid back and forward across factories.
Another new working paper by Hardy et al. (2024) explores one reason for this lack of labor fluidity across firms – information frictions. They are looking at self-employed garment makers, and interested in why more of them don’t merge together. They find 60% say they are willing to hire other firm owners to work for them, and 41% say they would be willing to work for a wage for another firm owner in the sector – yet this almost never happens. They find firms are willing to pay for information about which other firms are willing to hire, but then almost no mergers/working for another owner occurs, instead firms send some of their more senior apprentices over who are looking for jobs.
However, while it can be hard to get existing self-employed firms to merge, another recent working paper by Bassi et al. argues that some types of larger manufacturing end up looking like they are mergers of self-employed workers, since there is very little specialization within the firm and each worker within the firm does similar tasks to one another. This is again looking at bespoke manufacturing, like carpentry and welding. But I think what matters here is a lack of management and marketing skills, and building of brands. After all, we see large hair salon chains, plumbing companies, tax preparation firms, etc where most workers do the same jobs that are customizing solutions slightly for clients, but where firms have invested heavily in brand, customer acquisition, and back-office functions. This brings us to another firm boundary issue: when does the owner try to do everything for themselves on management, versus hire specialists? My work in Nigeria on the boundary of the entrepreneur, summarized in this post, looks at this issue.
Fluid labor can come with worker vulnerability
Typically economists think of firms as being large and risk neutral, and providing some forms of insurance to poorer and risk averse workers through the contracts they offer (e.g. workers paid a fixed salary are implicitly receiving insurance against the ups and downs of business demand). But with small firms in developing countries, we sometimes see that workers are instead the ones providing an insurance valve for firms. Firms may implicitly borrow from workers by delaying payments until they receive payment from a customer or suppliers (a frequent complaint of workers during some of my work in Nigeria), and firms may find labor much easier to adjust than capital, and so cut workers or wages when demand is low.
There doesn’t seem to be a lot of systematic documentation of this yet, but some evidence is starting to emerge from the Small Firm Diaries project. By collecting high frequency data on firms in multiple countries, this work is uncovering some of the volatility in work in small firms. Early results shared by Tim Ogden show that over 10 months, only 20% of workers in Nigeria and Colombia received pay for at least 8 of the 10 months. Moreover, even if employment is stable, wages need not be: they find pay can vary by 30% from month to month, even for long-term workers.
What about retail and services and non-bespoke manufacturing?
A lot of this work on capital fluidity seems specific to a few bespoke industries. Hardy et al. note they do not see any evidence of workers contributing capital in retail, food and handicrafts, nor in manual labor intensive work such as masonry, nor in skilled trades with costly machinery such as electricians. Rental markets for large capital equipment apply a bit more generally (e.g. tractors in agriculture), but likely account for a much lower share of total capital as one starts looking at firms operating at larger scale, operating production lines to standardize products, etc. But using temporary labor to adjust the workforce seems widely prevalent, and it seems some larger firms (and even governments!) may deal with liquidity issues by delaying paying workers. So the question of “what is a firm” may differ if you are looking at capital versus labor, as well as depending on which sector you look at. This is a new empirical literature, so I’m sure we’ll be seeing more insights to make us question what we think we know.
Join the Conversation