Why these slides?

Slides in first session of each pair:

​

• Give an overview of where I think the frontier is
• Note that this isn’t a guide of what to work on – indeed the most interesting questions may be missing because no-ones working on them!
• Instead I’m trying to put some framing on all the work currently going on in these (large) fields

​

• Not a literature survey – I try to reference these and focus on papers that are more recent

​

• Focus on papers published in top-5 and ones I think might publish there
• This is where you should be aiming when first thinking of research ideas (but nobody has three top-5 papers in their PhD!)
• Of course, fads come and go, so one shouldn’t be too driven by this – but if no papers are being well published on a topic, it’s worth asking why…

​

1

Firms in developing countries

2

Why are people in developing countries poor?

Most people get most of their income from labour

• Answer is mainly “because marginal product of labour is low”

​

So, why is marginal product of labour low?

​

Two possible answers:

• Lack of capital (including human capital)
• Low TFP (Total Factor Productivity)

​

To answer these qusetions, we typically look at firms as our unit of observation (though Mckenzie, 2024, describes issues with this)

3

Is it lack of capital?

• Clearly a big part of the answer (i.e. see Karen’s lectures on human capital)

​

• Can we do more to get firms capital effectively?
• D’Andrea & Limodio (2023, MSci): High-speed internet improves credit and firm outcomes
• Bari ⓡ al. (2024, AER) & Jack et al. (2023, QJE): Asset-based microfinance can have large positive impacts
• Cordaro et al. (2025): Microfinance can be more effective when contracts are performance contingent

​

​

​

​

​

​

​

​

4

More capital is probably not enough

• But more capital is probably not the whole answer:

​

• Hall & Jones (1999, QJE) suggest < 50%
• Caselli (2005) surveys this `development accounting’ literature

​

• Alvierez et al. (2023, JPE) find 1/3^rd of variation in output per worker is explained by `firm-embedded productivity’

​

• Giorcelli & Li (2025) find Soviet capital transfers to China only had long-term impacts when accompagnied with « know-how » transfer

​

​

5

Why is TFP low in developing countries?

Two possible explanations:

1. There are not enough productive firms: Need to “upgrade” firms
2. Factors are misallocated: Need to move factors to productive firms

​

Three broad questions:

• How important is each explanation?
• Why do these problems arise?
• What can we do about it?

​

Much of the work on the frontier is trying to answer one of these questions

​

​

​

6

1. Upgrading firms

• Verhoogen (2023, JEL) provides an excellent review of firm-related lit, so there aren’t many more papers for me to mention!

​

• Technology adoption a key part – covered in another class (?)

​

• Monitoring:
• Kelley et al. (2024, AER & JMP)
• Houeix (2025, JMP) – shows that adoption of digitial payment technology increases when not used as a monitoring technology

​

​

​

​

​

​

​

​

​

7

Other ways to upgrade firms

• Management also plays a key part
• Akcigit et al. (2021, AER) and Hjort et al. (2022) find that developing countries delegate too little / have too few managers, substantially lowering productivity
• Macchiavello & Morjaria (2022) show foreign take-overs improves coffee mill performance because they are more successful at improving management
• In general we know little on why management is poor and how to improve it…

​

• Small firms may suffer from slack
• Walker ⓡ  al. (2025) show that increasing demand can increase output (e.g. multiplier) because small firms face integer problems

​

​

8

2. Within-sector misallocation

• Misallocation (of labour) across sectors is for another class (?)

​

• Focus of a lot of firm-level work is misallocation within sectors:
• Restuccia & Rogerson (2017, JEP) provide a good review – conclude is important (though there is uncertainty as to how much)
• Key idea – systematic distortions prevent resources from being allocated efficiently

​

• They distinguish between:
• Indirect studies – Look at dispersion of marginal products
• Direct studies – Measure the impact of a particular source of misallocation

​

​

9

Indirect studies of misallocation

• Historically, estimates of misallocation come through looking at variation in productivity in survey data
• E.g. Hseih & Klenow (2009) estimate Indian manufacturing TFP would increase by >100% if misallocation eliminated – much bigger than the US
• But, as described in Verhoogen, measuring productivity is difficult…

​

• More recent papers look at variation in response to input shocks:
• Bryan et al. (2024, AER): Effects of larger loans very heterogenous across firms
• Hughes & Majerovitz (2025, JMP) find evidence of large amounts of misallocation among Sri Lankan microenterprises randomly given cash

​

​

10

Direct studies of misallocation

Various causes of misallocation which recent studies look at, including:

​

• Regulation
• Favouritism
• Contract enforceability
• Market power

​

11

Regulation driven misallocation - specific

E.g. literature on size-based regulation:

• Bertrand et al. (2024, ECTA): Ability of large firms to avoid regulation through contract labour reduced misallocation, increasing TFP by 8%

​

Some literature on tax audits:

• Henning & Okello (2024, JMP) show that comprehenseive tax audits increase distortions AND reduce government revenue

12

Regulation driven misallocation - informality

• Ulyssea et al. (2025) provides a good survey on informality

​

• Somewhat of a tension between impact evaluations (typically find small effects of formalisation policies) and general equilibrium estimates
• Wiseman (2023, JMP) actually finds big impacts on information-based RCT when focusing on informal cross-border traders

​

Recent JMPs look at informality in slightly different ways:

• Zarate (2024, JMP)
• Ramos-Menchelli & Sverdlin-Lisker (2022, JMP) find that increasing formal entry barriers can be welfare increasing by reducing inefficient firms entering

​

​

​

​

13

Misallocation driven by “favouritism”

Political connections:

• Ryan (2020, Ecta) shows connected Indian firms more likely to win bids unfairly – and drives misallocation as they are not lowest cost

​

Ethnic favouritism :

• Hjort et al. (2025, MSci) show co-ethnic investing very important on Kenyan Stock Exchange
• Fisman et al. (2020, JPE) show lending from Hindu bank managers to Muslims lower when managers have experience of riots

​

Perception of favouritism:

• Colonelli et al. (2024) show that correcting firms perceptions about government corrutpion increases their bids for public contracts – and the number they win

14

Misallocation driven by contract enforceability

Developing countries frequently have weak court systems making contract enforceability difficult. But courts don’t allocate resources, so why would this lead to misallocation?

​

• Rao (2025, JMP) finds that filling a judicial vacancy improves local firms access to credit, factor-use and ultimately production

​

• If contracts depend on relationships, firms better able to support relationships end up with larger market share
• See Macchievello (2022) for a survey of literature on relational contracts and development
• Wiles & Houeix (2025) show that randomized intervention to increase trust in trading relationships leads to sustained firm performance

​

​

​

15

Market power driven misallocation

If firms’ market power varies, marginal revenue ≠ marginal product – resources skewed to firms with market power

• Interesting to know extent and impacts of market power

​

• Rubens (2023, JMP, AER) looks at impact of an increase in monopsony power and markdowns in China, increased misallocation

​

​

​

16

A backlash to misallocation?

Existing studies may over-emphasise the importance of misallocation…

• Haltiwanger et al. (2018) – Indirect approach very sensitive to mispecfiifcation
• Bassi et al. (2022, Ecta): Effective rental markets mean misallocation of capital ownership isn’t as big of an issue
• Carillo et al. (2023) estimate extent of misallocation using procurement lotteries in Ecuador – find misallocation very small
• Gupta (2024, JMP): Part of estimated misallocation is demand-led – i.e. bigger firms have higher markups because of demand elasticity

​

​

​

​

17

Why so much interest in misallocation?

• Increasing factors of production is expensive… reallocating them seems easier

​

• Particularly if simply involves removing distortive policies!

​

• Still the case that wins are large if we can identify such cases

​

• But increasingly seems we may have to reallocate across sectors / firms in a more active manner…

​

​

18

Industrial policy

If we can’t remove the distortions, should we create compenstating distorsions – i.e. industrial policy?

​

• Juhasz, Lane and Rodrik (2024, ARE) and Reed (2024, JEP) are good surveys

​

• Many new papers on industrial policy, e.g.:
• Garg (2025, JMP) looks at it’s role in solving coordination failures
• Wang & Xing (2025, JMP) find EV subsidies in China had no effect on welfare because of entry of « lemons » and hence negative reputation effect
• Barwick et al. (2024) compare different sorts of industrial policy to support EVs

​

There has been a taboo to working on this until recently – so lots of unanswered questions - i.e. How exactly should we do industrial policy?

19

An important note

Not all papers working on firms look at upgrading or misallocation, e.g.:

• We are interested in how much tax firms pay from a government revenue perspective
• E.g. Carrillo et al. (2023, AERI) look « ghost firms » and how they are used
• We are interested in market power for other reasons:
• Bergquist & Dinerstein (2020, AER) show traders capture almost all the surplus
• Grant & Startz (2025) show shortening supply chains can reduce welfare by increasing market power
• Aspects of firm behaviour that don’t fit neatly into these boxes:
• Banerjee et al. (2024) show in that firms don’t maximize profit even when shown how to…

​

See BREAD-IGC Virtual Course for more!

​

​

​

20

A thought on working on firms & development

Yes there is opportunism, but ideas are also very important:

​

• Note that some of the best papers are applying an idea to a recent dataset / ID strategy that hadn’t been fully exploited
• Hence importance of following what’s at the frontier

​

• Note that the firms studied are often quite random
• Paucity of detailed data and importance of measurement means there may be opportunities working with any firm(s)
• E.g. Startz (2025, JMP)

21