Large income gaps across countries remain one of the most pressing issues in global economic policy. A long-standing view among economists is that these gaps mainly reflect differences in productivity rather than differences in capital or labour alone (Klenow and Rodríguez-Clare 1997, Prescott 1998, Jones 2016). Over the past decade, attention has increasingly shifted to how productivity is shaped by the way resources are allocated across firms. Influential research building on Restuccia and Rogerson (2008) and Guner et al. (2008), most notably Hsieh and Klenow (2009), shows that in many developing countries, productive firms remain too small, while less productive firms are too large (see Hopenhayn 2014 and Restuccia and Rogerson 2017 for surveys). This misallocation of resources can lead to large losses in aggregate productivity. Yet there is still limited cross-country evidence on the extent of misallocation and on which features of the business environment are most responsible for distortions creating misallocation.

This column summarises new evidence from my paper (Restuccia 2025) using the recently expanded World Bank Enterprise Surveys (World Bank 2025), which cover firms in more than 100 countries in manufacturing and services. The analysis applies a standard and widely used framework of business size to measure firm-level productivity and the distortions implied by firms’ observed size.

The main message is not just that distortions are larger in poorer countries, but that they operate differently. In developing economies, distortions are more closely linked to firm productivity, weakening the expected connection between how productive a firm is and its operational scale. As a result, the productivity costs of misallocation are much higher. The World Bank data also allow an exploration of which obstacles faced by firms – such as regulation, finance, corruption, or weak infrastructure – are most consistent with the empirical pattern of distortions.

How productivity and distortions are measured

The analysis builds on a simple idea based on standard models of business size (Lucas 1978, Hopenhayn 1992). Firms differ in how productive they are, and in an economy without major distortions, more productive firms should employ more workers and operate at a larger scale. This creates a strong positive relationship between firm productivity and firm size. When this relationship breaks down, it signals the presence of distortions – such as regulations, informal payments, or other barriers – that affect firms’ ability to reach their efficient size.

Using the World Bank data, I measure firm productivity from information on sales and employment. Distortions are inferred from how far a firm’s size deviates from what would be expected given its productivity. To make comparisons meaningful across countries, industries, and survey years, I focus on relative differences across firms within each country rather than absolute levels.

The World Bank Enterprise Surveys are particularly useful for this purpose because they apply a common survey design across a large number of countries and include detailed questions about firms’ operating environments. At the same time, the data have limitations. They exclude farm businesses that are much more prevalent in developing countries and may be subject to distinct barriers and distortions than those characterised by the Enterprise Surveys. The data also exclude small firms (those with fewer than five employees), and sample sizes tend to be relatively small, which prevents more granular comparisons across countries. Nevertheless, the analysis explicitly checks that these sample features do not drive the main results.

Larger productivity gaps in poorer countries

The first set of results concerns differences in productivity across firms (Figure 1). Firm-level productivity is much more uneven in developing countries than in advanced economies. Importantly, this is not because developing countries have a few extremely productive firms. Instead, they have many very low-productivity firms. The bottom of the productivity distribution is much lower, creating a wide gap between the most and least productive firms. These findings are consistent with previous studies from a limited set of countries and Ayerst et al. (2024), who used Orbis data.

Figure 1 Productivity gaps across firms and level of development

Notes: Productivity dispersion in each country is measured by the standard deviation of log total factor productivity (TFP) across firms in panel (a) and the indicated percentile ratios of firm TFP in panel (b). Each observation is the value for the indicated country. Labour productivity in logs for the year 2023 from the Penn World Table v11.0 (Feenstra et al. 2015).

These large productivity differences make efficient allocation especially important. When productivity varies widely across firms, the gains from reallocating employment toward more productive firms are potentially large. This is precisely where distortions matter most.

A weaker link between productivity and firm size

My central empirical finding is that distortions in developing countries weaken the relationship between firm productivity and firm size. In richer economies, more productive firms are consistently larger. In poorer economies, this link is much weaker: highly productive firms remain relatively small, while less productive firms absorb more workers than they should.

This pattern can be seen directly in the data. The responsiveness of employment to productivity – the extent to which firms attain a larger size as they are more productive – is much lower in developing countries. At the same time, distortions faced by firms are not only more dispersed, but also more strongly related to productivity. Productive firms in poorer countries tend to face higher effective barriers to their operating size (Figure 2).

Figure 2 The link between distortions (employment) and productivity across firms

Notes: The relationship between distortions and productivity across firms in panel (a) and employment and productivity in panel (b), both measured as the elasticity of log distortions (employment) on log productivity across firms in each country. Labour productivity in logs for the year 2023 from the Penn World Table v11.0 (Feenstra et al. 2015).

This combination is particularly damaging. Distortions that fall more heavily on productive firms prevent factor inputs such as employment from moving to where they are most efficiently used, generating large losses in aggregate productivity.

How large are the productivity losses?

To quantify these effects, I compare actual aggregate output to the output that would be achieved if labour were allocated efficiently across firms within each country. This measure, often referred to as allocative efficiency, captures how much productivity is lost because workers are not employed where they are most productive.

The results show that allocative efficiency is much lower in developing countries (Figure 3). In many cases, actual aggregate output is less than half of what could be achieved with a more efficient allocation of labour. These losses are substantial in both manufacturing and services, but they tend to be larger in services, where regulation and non-tradability are more common.

Figure 3 The productivity cost of misallocation within manufacturing and services

Notes: Allocative efficiency is the ratio of aggregate actual relative to efficient output. Labour productivity in logs for the year 2023 from the Penn World Table v11.0 (Feenstra et al. 2015).

These findings are consistent with earlier studies for individual countries, but the World Bank Enterprise Surveys data make it possible to document them systematically across a wide range of economies using a common approach.

Which obstacles matter most?

A key advantage of the World Bank Enterprise Surveys is that they include firms’ own assessments of the obstacles they face. This makes it possible to examine not only which constraints are more common in developing countries, but also how these constraints vary across firms within a country.

My analysis shows that several aspects of the business environment are especially consistent with the observed pattern of distortions. Regulations that affect a firm’s prices, for example, are much more common in developing countries and tend to affect more productive firms more strongly (top panels in Figure 4). Measures of corruption, such as the frequency of informal payments, show a similar pattern. Infrastructure problems, including internet and electricity disruptions, also disproportionately affect productive firms in poorer economies.

Figure 4 Regulation and finance and level of development

Firms reporting their prices to be regulated

Firms reporting that they are credit unconstrained

Notes: The top panels refer to a measure of regulation (firms reporting their prices to be regulated) and the bottom panels refer to a measure of finance (firms reporting that they are credit unconstrained). The left panels report the average of each measure for each country, and the right panels, the difference in each measure of the average of firms in the top and bottom 10% of the productivity distribution. Labour productivity in logs for the year 2023 from the Penn World Table v11.0 (Feenstra et al. 2015).

By contrast, financial constraints tell a more nuanced story. While access to finance is clearly worse on average in developing countries, financial indicators do not consistently show that more productive firms are more constrained than less productive ones in a way that mirrors the measured distortions (Figure 4, bottom panels). This suggests that not all commonly cited obstacles play the same role in generating misallocation, at least on a systematic basis across countries. It could well be that some aspects of the business environment are more harmful in some sectors and in some countries than in others – an issue that would require more detailed country-specific analysis.

The distinction between the average distortion and how distortions differ across firms is important. Constraints that affect all firms similarly may reduce overall output because of their effects on capital accumulation or other inputs, but do not necessarily lead to severe misallocation. Constraints that vary across firms – and especially those that penalise productive firms – are far more harmful for aggregate productivity.

Policy implications

These findings have clear implications for policy. Improving productivity in developing countries requires more than reducing average barriers to doing business. It also requires addressing how those barriers affect firms differently. Policies that reduce uneven regulatory burdens, corruption, and infrastructure failures – particularly where they constrain productive firms – can deliver large productivity gains.

The results also highlight the limits of relying solely on country averages when assessing the business environment. Two countries with similar average levels of regulation or corruption may experience very different aggregate productivity outcomes if these constraints fall more heavily on their most productive firms.

Overall, the evidence suggests that misallocation remains a central reason why productivity is low in many developing countries. But the problem is not simply that firms face many obstacles. It is that these obstacles systematically weaken the link between productivity and firm size. Restoring that link – so that productive firms can grow and attract more resources – should be a central goal of productivity-enhancing reforms.

References

Ayerst, S, DM Nguyen, and D Restuccia (2024), “The micro and macro productivity of nations”, NBER Working Paper 32750.

Feenstra, R C, R Inklaar, and M P Timmer (2015), “The next generation of the Penn World Table”, American Economic Review 105(10): 3150–82.

Guner, N, G Ventura, and Y Xu (2008), “Macroeconomic implications of size-dependent policies”, Review of Economic Dynamics 11(4): 721–44.

Hopenhayn, H A (1992), “Entry, exit, and firm dynamics in long run equilibrium”, Econometrica: Journal of the Econometric Society 1127–50.

Hopenhayn, H A (2014), “Firms, misallocation, and aggregate productivity: A review”, Annual Review of Economics 6(1): 735–70.

Hsieh, C T, and P J Klenow (2009), “Misallocation and manufacturing TFP in China and India”, The Quarterly Journal of Economics 124(4): 1403–48.

Jones, C I (2016), “The facts of economic growth”, in Handbook of Macroeconomics, Vol. 2, Elsevier.

Klenow, P J, and A Rodriguez-Clare (1997), “The neoclassical revival in growth economics: Has it gone too far?”, NBER Macroeconomics Annual 12: 73–103.

Lucas, Jr, R E (1978), “On the size distribution of business firms”, The Bell Journal of Economics 9(2): 508–523.

Prescott, E C (1998), “Lawrence R Klein Lecture 1997: Needed: A theory of total factor productivity”, International Economic Review 39(3): 525–51.

Restuccia, D (2025), “On productivity and distortions across countries”, NBER Working Paper 34573.

Restuccia, D, and R Rogerson (2008), “Policy distortions and aggregate productivity with heterogeneous establishments”, Review of Economic Dynamics 11(4): 707–20.

Restuccia, D, and R Rogerson (2017), “The causes and costs of misallocation”, Journal of Economic Perspectives 31(3): 151–74.

World Bank (2025), Enterprise surveys.