Setting up your own firm for a firm experiment


This page in:

The typical approach to examining how workers, consumers, or governments interact with a firm has been for researchers to find a willing firm owner and convince them to run experiments. Examples include Bandiera et al. working with a UK fruit-farmer to test different payment incentives for immigrant workers; Bloom et al. working with a Chinese travel agency to test the effect of letting workers work from home; and Adhvaryu et al. working with an Indian garment firm to measure impacts of soft-skills training for workers and of introducing LED-lighting. However, finding/persuading a firm to do the experiment that a researcher would like to do can be hard, with many of these existing samples coming about through a researcher having a former student or relative who runs one of these firms.

So what should you do if you lack a connection, or you want to do something that you cannot persuade a firm to do?

Recently, a number of researchers have taken a different approach, which is to set up and run for themselves a firm in order to answer research questions. I thought I would give some examples of this work, and then discuss some of the issues that arise or things to think about when deciding about pursuing this research strategy.

Recent examples of researchers setting up their own firms to use for research:

  • Firm-customer relationships: Abhit Bhandari (on the job market) created and registered a legal business, and hired 9 employees to sell a mobile phone credits at a discount door-to-door directly to households in Senegal. The aim was to examine how political connections moderate economic transactions that involve a moral hazard component, where delivery takes place after payment. He then randomized whether employees signaled their (author-arranged) connections to local government, and/or offered a formal contract at the time of trying to make a sale.
  • Firms and gender: In exciting work in progress, Solène Delecourt and Odyssia Ng (both PhD students at Stanford), set up their own market stalls selling vegetables in India, and randomly vary the gender of who is running the stall in order to better understand the reasons behind gender profit gaps.
  • Firm-worker relationships: Emily Breza, Supreet Kaur, and Yogita Shamdasani set up their own factory workshops in India, and employed 378 workers full-time for one month in seasonal jobs to manufacture local products such as rope, brooms, and floor mats. They then randomize whether workers within a production unit all get the same pay, or whether they get paid differentially according to baseline productivity to test how pay inequality affects output and attendance.
  • Firm-funder relationships: In work forthcoming in the next couple of months, Suresh de Mel, Chris Woodruff and I hired a lawyer in Sri Lanka to understand how one could make micro-equity investments in microenterprises, and then de Mel was involved in setting up a company that could enter into partnership agreements with selected firms. We ran this as a proof-of-concept to examine how micro-equity could function.
I should also note that an intermediate approach between trying to persuade another firm to answer some question for you, and setting up your own legal entity to do so has been for researchers to use their survey firm or research institute to be the employer for labor experiments. Examples of this include Beaman and Magruder and  Marc Witte (also on the job market) who both look at job referrals for short-term jobs they set up.

Practical issues to consider when doing this:
Setting up your own firm can enable researchers to test specific mechanisms that may not be possible to do within an existing firm, and, as with fieldwork, the very process of setting up a firm and dealing with the practical logistical issues involved may give researchers new ideas and impart lessons that would be difficult to learn from just theory and data. However, there are a number of practical issues to consider when doing this:
  1. Is this legally feasible? Some countries make it very hard for foreigners/non-residents to set up a firm, and even when legal, it may be expensive and time-consuming. Bhandari reports it took approximately one month from start to finish to set up a business under Senegal’s simplified “one-stop shop” regime. We had to hire a lawyer in Sri Lanka to explore how to best set up micro-equity contracts and the appropriate legal vehicle for doing this. In contrast, if what you are studying is an informal vendor, full registration is not needed. But even then, it may be non-trivial – Delecourt and Ng report to me that they rented stalls in designated markets from market leaders, but that it was sometimes difficult to find and meet with these market leaders, and that sometimes the government would shut down the entire market to use the land for other purposes, which also meant shutting their businesses.
  2. How long does your company need to last, and what will you promise your workers? Most of the companies set up have been very temporary, and explicitly offered very short-term contracts to workers. E.g. Bhandari hired workers for 2 months; Breza et al. have their company produce for a month only; Delecourt and Ng recruit vendors for a day at a time; and Witte hires workers only for 3 days. This can work fine for short-term experiments about worker productivity, but much longer time frames are needed to answer other questions – in Sri Lanka, we began trying to set up contracts in 2012, and then had to wait for investment pay-back periods of several years, continuing through 2017.
  3. Profit or non-profit-making enterprise? Several considerations arise here. One is whether funders will want to support research that involves supporting a profit-making firm, a second is what happens to the profits (do they go to the researchers? Are they donated? Etc.) A third issue is a potential ethics/optics concern with a firm set up by researchers making money from poor customers in a developing country. A fourth is a potential conflict-of-interest that can arise from researchers undertaking experiments in a company they have a financial interest in. For these reasons, most of the examples above do not try to make profits (or end up making losses). This contrasts with firms set up by researchers like the Entrepreneurial Finance Lab, and StickK, in which the researchers behind them decided not to conduct academic experiments in their own firms.
  4. Impacts on other businesses in the economy?  This may especially be an issue with non-profit-making businesses – deliberately making a loss in order to get enough sales may displace sales from other businesses, while paying workers more than prevailing wages may make it harder for other firms to hire workers – so be careful to minimize these impacts.  Delecourt and Ng tell me that they did sometimes get protests from other vendors against their entry into the market (more because of their outsider status than because of an increase in competition), forcing them to move from some markets.
  5. Realism/Generalizability: depending on the question being asked, there may be a concern that companies set up by researchers are somehow less realistic or the results from them may be less generalizable than those from a real profit-making firm operating for a non-research purpose.
  6. Liability issues/risk of harm? Once researchers start hiring workers as employees for their firm, they may then face legal liability if anything happens to them in the workplace. One of the researchers notes to me that hiring employees was a lot more complicated from an IRB perspective than the standard hiring of enumerators for survey purposes, and they needed to use a lawyer to ensure local labor laws were followed. Similarly, there may be customer complaints and liability to deal with. We might be fine with researchers setting up their own data entry firms or having them sell phone credit, be a little more worried about manufacturing firms, and be even more concerned if we see researchers setting up their own health clinics or schools – although of course they may well offer a service much better than the counterfactual.
  7. A lot of local knowledge may be required to do this successfully: while your purpose in doing this may be to avoid the need for close connections to a firm, often you will need to know a lot about local context and may need local connections to make your business work.
  8. Running a business is hard work, and there are lots of logistics to deal with! Delecourt and Ng gave me several examples, such as “We had to buy hundreds of pounds of vegetables from the wholesale market, which closes around 6 am. Therefore, a team member had to go buy vegetables in the middle of the night”; “Figuring out which products to get, at what price, and in what quantity was a challenge, especially as vegetables are seasonal”, and “Another problem we faced are animals that would come eat our leftovers overnight”


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

Rem Koning
January 22, 2019

Setting up a new "division"
An in-between approach can sometimes be effective. Instead of starting a new firm or trying to tinker with an existing company, you help an existing organization/firm launch/develop a new "division" where you can "naturally" build in randomizations and measurement. It also means the company is more likely to try something new since there isn't an existing way of doing things. This sort of change is often the hardest part of the process when working with firms.
With Solene Delecourt (see above), Sharique Hasan (Duke), and Ronnie Chatterji (Duke) we worked with a leading Indian business association to design a new executive retreat for their members. The design of the retreat included peer randomizations and surveys designed to help attendees benchmark their startups. I think the fact that we helped with the design of a new program allowed us to do these two things. Using this data, we find peers are valuable sources of management advice impacting firm outcomes two-years after the retreat, but only when a founder doesn't have an MBA and is not in a startup accelerator. We think this is evidence for a tradeoff between formal training and informal knowledge sharing. The paper here: I
With Sharique Hasan we founded a startup bootcamp with IIIT-Delhi. Papers still under review, but we learned a lot from this. The first thing we learned is that getting an adequate sample size is tremendously hard. These sorts of "new" programs often attract dozens to a few hundred participants/employees. Even when there is more demand, your partner might want to skim from the top to keep the program selective. Attrition is often meaningful. We ended up with 112 participants in the startup bootcamp. Enough to analyze some research questions, but was limiting and worth thinking through earlier than we did.
The second learning is one of duration, similar to bullet 2 above. The program was three-weeks long, and all our measurement happened during the bootcamp. In retrospect, I think it would have been better to design interventions during the bootcamp and track longer-term measures after the fact. Reviewers can always push back that outcomes are not long-term, which is something we experienced. Also, if you have kids or teaching obligations, running a program for longer than a month or two is often infeasible.
Finally, I want to add three names to the firm-founder+researcher list.
One is a Sociology/Strategy student on the job market our of HBS, Stefan Dimitriadis, started his own training program in Togo to test how different styles/cultures of communication impact knowledge sharing, the development of social captial, and firm performance.
The second is Vanessa Burbano at Columbia Business School who has built "firms" on UpWork to look at who applies to a firm that signals it engages in CSR.
Paul Niehuas of UCSD and GiveDirectly is also another model for this sort of work. He started GiveDirectly as a graduate student, and from my understanding, experimentation is built into the non-profit. A right tail outcome, but worth keeping in mind when contemplating the potential impact of starting your own venture.

David McKenzie
January 23, 2019

A couple of readers also alerted me to this example of researchers setting up their own temporary health clinic in Oakland to look at the effect of the Doctor's race on patient take-up of preventative care…
The paper does not provide much detail on the logistics of setting up a health clinc, but note that they ran the clinic on only 11 Satutdays, staffing it with 14 doctors and receptionists.

David McKenzie
January 23, 2019

And apparently setting up your own school is a thing too - Robert Metcalfe notes that John List set up his own pre-K school. From List's webpage "In 2009, Professor List, together with world-renowned scholars at the University of Chicago and Harvard University, was awarded a $10M grant from the Griffin Foundation to launch the Chicago Heights Early Childhood Center (CHECC), one of the most comprehensive longitudinal early childhood studies ever conducted. With nearly 1,500 child participants, CHECC aimed to identify programs and strategies for improving student education in America’s schools and to try to understand how to best foster development during the early years of a child’s life and to place children on a successful path. "