Should governments aiming to improve job opportunities devote additional resources towards trying to provide programs that attempt to generate marginal changes in many micro and small firms, or try to target the support towards making larger impacts on a smaller number of high-growth and larger firms? For example, should a government spend an additional $5 million on grants and training programs that support 25,000 micro firms at $200 each, use it to give 100 grants of $50,000 each to 100 high-growth potential firms, or use it as a single $5 million tax incentive to encourage one large multinational to set up a manufacturing plant in the country? I’ve been asked my thoughts on this question quite a few times, so thought I’d share them here.
The answer involves many different trade-offs and considerations, and I attempt to summarize some of the key ones in this post. The bottom line is that there are trade-offs (at least in the short-run) between poverty alleviation and productivity growth, and that different policies will have impacts on different types of job creation. A key lesson for policymakers is to be clear about what the job problem is that they are trying to solve, and not try to use the same policy instrument to achieve multiple competing priorities.
One standard reaction I have heard is for people to refer to research on whether small vs large firms, or young firms create more jobs (e.g. Ayyagari et al, 2014). However, while useful as a descriptive of where jobs are being created and destroyed, such studies tell us very little about policy effects. What matters are the policy deltas – the additional jobs created if additional funds are devoted to helping a particular type of firm compared to business as usual. For this, explicit consideration of counterfactuals is needed.
The case for (and against) policies targeted at many micro and small firms
Micro, small, and medium firms are important sources of employment for the poor. Successful finance, grant, and training programs directed towards such firms typically improve self-employment prospects for the owners of these firms, helping them to start such enterprises, have these enterprises survive longer, and help their owners earn higher incomes. However, they have little discernable effect on paid employment of others (Grimm and Paffhausen, 2015). But these programs can have lasting impacts on the incomes of relatively poor people. For example, in de Mel et al. (2012) my co-authors and I show that one-time grants of $100 to $200 given to Sri Lankan male business owners resulted in 10-percentage-point-higher enterprise survival rates, and $8-to-$12-per-month-higher profits five years later, resulting in substantial returns to capital. Likewise, two recent evaluations I have done of business training programs in Kenya and Togo find limited employment effects, but significant increases in the incomes of firm owners that more than result in high return on investments (McKenzie and Puerto, 2017; Campos et al, 2017).
In absolute terms, these changes for each firm served by these programs are small. But an increase in income of $8 to $12 per month in Sri Lankan firms that were earning $47 per month, or an increase of $2.60 per week in Kenya for women who were earning $13 per week equates to an important improvement in the returns to the labor of the poor. Moreover, in the Kenyan study, we are able to show that this increase in income does not come from reducing the income of other businesses operating in the same markets, but that it comes from growing the market as a whole.
A final point in favor of policies targeted at such firms is that emerging evidence suggests that this is the group of firms that policymakers may be best able to predict success amongst, at least in terms of distinguishing between firms with growth prospects and subsistence firms (e.g. Hussam et al, 2017). Moreover, several of the evaluations show both high returns for the average firm, and impacts across a large fraction of the distribution.
The case against such policies is that such policies may be much better at poverty alleviation than at generating long-term productivity growth and structural change in the economy. Very few of these microenterprises ever grow to the point of hiring workers, and especially to hiring more than ten workers (see background note on targeting). Moreover, as economies grow, the share of the labor force in self-employment shrinks dramatically (Gollin, 2008), and so efforts to help the microenterprise sector are largely focused on firms that we expect to eventually disappear, and on individuals who we eventually expect to become wage workers, as the economy grows. A final point to note is that it can be costly to administer a program to thousands of firms, and so a larger share of my hypothetical $5 million in spending might end up getting spent on program administration costs than in programs helping larger firms.
The case for (and against) policies targeted at high-growth firms
High-growth firms account for a disproportionate share of employment growth, and so successful policies which are targeted towards such firms can generate new jobs in the economy. For example, in McKenzie (2017), I show that individuals who win a $50,000 grant in a business plan competition in Nigeria are more than 20 percentage points more likely to have grown their firm to 10 or more workers three years later than comparable non-winners, with a cost per job created that is substantially cheaper than is the case for vocational training, wage subsidy, management training, and small grant interventions. Likewise, Fafchamps and Quinn (2017) find short-term positive and cost-effective impacts on job creation of a business plan competition in three other African countries.
The types of jobs created by these high-growth firms appear to be somewhat different than those in micro-enterprise employment. In my work on the Nigerian business plan competition, I find that the workers hired by the business plan competition winners tend to be more educated than the average Nigerian youth, and such jobs may do less for poverty alleviation than directly helping the self-employed. However, these jobs may be ones in which workers are more likely to learn new skills, and which may contribute more to productivity growth in the country.
Two disadvantages of such policies are the difficulty in targeting such programs, and the cost. In work with Dario Sansone, I note that it has proven very difficult to predict in advance which firms will grow rapidly. The best that may be able to be done is to screen out the bottom tail of firms, and use a mechanism which self-selects a pool of firms with more growth potential. Secondly, the cost of these programs may make it expensive to reach very many firms. The YouWin! business plan competition spent $60 million on grants to 1,200 firms in order to generate just over 7,000 jobs. While cost-effective compared to many alternative job policies, this still highlights the high cost of generating jobs through direct support to high-growth firms.
A further trade-off can arise between government efforts to support innovation and the development of high-technology industries, versus a desire to support job creation. Programs that improve the productivity and innovativeness of digital start-ups, biotechnology, etc. may create the types of high-paying, high-skilled jobs that are scarce in many developing countries. But these sectors also tend to not be very employment-intensive, and support to other sectors may be more effective in generating large numbers of job for poorer and less-skilled individuals. In some cases, supporting productivity improvements in these mid-sized firms may actually reduce employment in the long-term, as seen in my long-term follow-up of a management improvement intervention in India.
The case for (and against) supporting very large firms
A final alternative is to focus very narrowly on trying to attract more large firms to set up operations in the country, and policy actions to help the largest firms grow. Some of the policies to do this are more general business environment and infrastructure policies that may help firms of many different sizes. The broader trade-off question arises when considering direct efforts to attract and help the largest firms. These may include tax incentives, free trade zones, research and development incentives, supplier development programs, and other such policies. The argument in favor of such efforts is that these large firms may generate many jobs at once, and help improve the productivity and industrial capacity of a country.
However, there are several reasons to be cautious about the ability of such policies to generate employment. The evidence base on many of these interventions is sparser and less well-identified than on interventions for microenterprises and high-growth firms. One partial exception is work on export processing zones. Cirera and Lakshman (2017) conduct a systematic review of 59 studies and conclude there is no robust evidence that these zones create additional employment. A second concern is one of political capture and fairness, with incentives ending up largely benefiting well-connected and already wealthy business owners in many developing countries. Rodrik (2004) lays out a set of principles that an appropriate industrial policy should follow, but, to my knowledge, there is no rigorous evidence as to the extent to which attempting to follow such policies has resulted in employment growth.
- Generating additional employment is hard to do, and hard to measure. Many impact evaluations are only short-term, and are unable to measure whether new jobs are created, or whether firms supported take away business from other firms, and whether the workers employed simply move away from other employment activities. Context matters here – countries with high unemployment rates and/or large numbers of youth entering the labor market every year may be more focused on the number of jobs created and less on displacement, while those with high employment rates may instead be trying to create more jobs of higher wages and quality.
- Governments are often guilty of trying to use the same policy instrument to try to serve multiple policy aims that might conflict with one another. For example, the appropriate policy instrument to spur high-growth entrepreneurship is unlikely to also be the right instrument for including under-served groups such as underrepresented regions, female entrepreneurs, or the poor. Policies to encourage innovation and the development of a high-tech sector may be more effective if they are not also trying to select firms on employment potential. Governments should therefore be very clear what job problem they are trying to solve before deciding which policy instrument will be most successful for doing so.
- While the choice of where to spend time and money at the margin will involve a trade-off between supporting microenterprises, SMEs, and large firms, most countries will want a mix of policies that offer different policy instruments designed to support all of these types of firms. Indeed, we see this is still the case in most developed countries, where governments both spend money on small-business programs, as well as providing tax incentives and other programs for large firms.
- A final point is that I am sympathetic to Lant Pritchett’s point that such money might be even better spent on something with only a tiny probability of having an influence, and with it being difficult to ascribe causality, but where the upside potential is huge.