3

CHAPTER
SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE
TO TURN THE TIDE?

The global economy, while demonstrating remarkable (Kilic Celik, Kose, and Ohnsorge 2023). This suggests
resilience to recent shocks, faces a sobering reality: its a possible downshift to a lower-growth regime.
medium-term growth prospects have consistently been revised The growth decline implies worsening prospects
downward since the 2008–09 global financial crisis. This for living standards and global poverty reduction. An
reflects a downward trend in actual global growth, with the entrenched low-growth environment, coupled with high
slowdown starting in the early 2000s in advanced econo- interest rates, would threaten debt sustainability and
mies and after the crisis in emerging market and developing could fuel social tension and hinder the green transition.
economies. This chapter examines the factors behind this Furthermore, expectations of weaker growth may deter
trend, revealing that a significant and broad-based slowdown investment in capital and technologies and so, in part,
in total factor productivity growth accounted for more than become self-fulfilling. Therefore, addressing the weaken-
half of the growth decline. This deceleration was driven in ing growth outlook is a policy priority for all economies.
part by increased misallocation of capital and labor across Changes in growth performance can be attributed
firms within sectors. A widespread drop in postcrisis private to the contributions of labor and capital inputs and
capital formation and slower working-age-population the efficiency of their use—known as total factor
growth in major economies exacerbated the slowdown. This productivity (TFP). Among these proximate drivers,
chapter predicts that, without timely policy interventions or growth in labor inputs is held back by demographic
a boost from emerging technologies, global growth will be pressures and declining labor force participation trends
only 2.8 percent by the end of the decade, significantly below (Chapter 2 of the April 2018 World Economic Outlook
its prepandemic (2000–19) average by a gap of 1 percent- [WEO]; Goodhart and Pradhan 2020). In addition,
age point. This highlights the urgent need for policies and ever since the global financial crisis, anemic private
structural reforms that enhance growth by improving capital investment in advanced economies has impeded
and labor allocation to more productive firms, enhancing capital deepening (Chapter 4 of the April 2015 WEO;
labor force participation, and harnessing the potential of Döttling, Gutiérrez, and Philippon 2017). However, a
artificial intelligence. Such measures are critical, especially in comprehensive analysis of business investment dynam-
light of challenges such as high public debt and geoeconomic ics that includes emerging market economies is lacking.
fragmentation, which could further constrain global growth. TFP, a prime contributor to trend growth, can
increase through within-firm productivity increases
resulting from technological progress and through
Introduction better resource allocation across firms—resources flow
Since the 2008–09 global financial crisis, forecasters toward more productive firms—improving overall
have persistently lowered their expectations for “allocative efficiency” in an economy (Restuccia and
growth over the medium term (Figure 3.1). Estimates Rogerson 2008). Whereas technological advances
of potential output growth—an economy’s maxi- have attracted extensive research, little attention has
mum noninflationary growth given its resources and been paid to how allocative efficiency varies over time
technological capabilities—indicate a similar decline and how shifts in allocative efficiency have affected
TFP growth.1 To fill this gap, this chapter employs an
The authors of this chapter are Nan Li (co-lead), Chiara Maggi,
Diaa Noureldin (co-lead), Cedric Okou, Alexandre B. Sollaci, and
Robert Zymek, with support from Shrihari Ramachandra, Pablo 1The contribution of slowing innovation to the decline in TFP

Vega, Yarou Xu, and Dennis Zhao. The work in this chapter is partly growth has already been studied extensively; see, for example,
supported by the Macroeconomic Policy in Low-Income Countries Gordon (2016); Bloom and others (2020); Chapter 3 of the October
program of the UK’s Foreign, Commonwealth and Development 2021 World Economic Outlook; and Acemoglu, Autor, and Patterson
Office (FCDO) and the Macroeconomic Research on Climate Change (2023). In addition, a large body of literature, surveyed in Restuccia
and Emerging Risks in Asia program of the Ministry of Economy and and Rogerson (2017) and including Chapter 2 of the April 2017
Finance of the Government of Korea. The views expressed do not nec- Fiscal Monitor, has studied the role of misallocation in explaining
essarily represent the views of the supporting partners. Peter Klenow global gaps in productivity levels. Unlike that literature, this chapter
was the external consultant. The chapter benefited from comments by focuses on changes in misallocation over time, their causes, and their
Chang-Tai Hsieh and internal seminar participants and reviewers. contribution to recent and prospective TFP growth.

International Monetary Fund | April 2024 65

 WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

Figure 3.1. Five-Year-Ahead Real GDP Growth Projections, To answer these questions, the chapter begins by
2000–29 examining medium-term (five-year-ahead) WEO
(World growth, percent) growth projections, alongside actual growth trends,
over the past three decades across a wide range of econ-
5.0
omies. Subsequent sections provide in-depth analysis of
the proximate drivers of growth: labor inputs, private
4.5
capital formation, and allocative efficiency. Last, the
chapter presents various scenarios to assess the likely
4.0 growth paths in the medium term and the potential
effects of policy interventions.
3.5 The chapter’s main findings are as follows:
• The decline in medium-term growth projections
3.0 World Economic Outlook forecast is widespread, reflecting secular forces rather than
Consensus Economics forecast forecaster pessimism. Expectations for medium-term
growth have been revised downward across all
2.5
2000 04 08 12 16 20 24 29 income groups and regions, most significantly in
emerging market economies.
Sources: Consensus Economics; and IMF staff calculations. • Actual growth has similarly declined, and this is
Note: World Economic Outlook (WEO) sample comprises 196 economies and
Consensus Economics sample comprises 88 economies. Global real GDP growth largely because of TFP growth dynamics. In advanced
projections are calculated using GDP in purchasing power parity in international economies, productivity growth started to decrease
dollar weights. The years on the horizontal axis refer to the year for which a
forecast is made, using the April WEO from five years earlier. For example, the before the global financial crisis. In contrast, TFP
2029 forecast is based on the April 2024 WEO, and so on. The red line depicts the growth in emerging market and developing econo-
mean of the Consensus Economics forecasts.
mies rose before the crisis and then fell, mirroring
the globalization cycle. For both, changes in TFP
growth have significantly shifted overall economic
approach developed by Hsieh and Klenow (2009) that growth, accounting for more than half of the decline
proposes that a growing gap in revenue productivity in advanced and emerging market economies and
among firms signals a decline in allocative efficiency nearly all of the decline in low-income countries.
(see Box 3.1 for detailed explanations of the notion • Increased misallocation of capital and labor among
and measurement of allocative efficiency). firms has exerted a drag on TFP of 0.6 percentage
In this context, this chapter seeks to answer the point a year in the economies considered in the analysis.
following questions: This suggests that TFP growth could have been
• What are the insights from forecasts? How did fore- 50 percent higher if misallocation had not increased.
casters’ views on medium-term growth evolve, and Most of this misallocation increase is because of
what do they imply about income inequality and uneven firm productivity growth within sectors,
convergence? requiring reallocation of capital and labor, which was
• How did we get here? What factors account for the impeded by economic frictions. Although shocks
decline in actual growth over the past two decades? may temporarily worsen misallocation, two-thirds of
What role did demographics and private investment it at any time can be attributed to persistent struc-
play? To what extent have changes in allocative tural frictions, which policy measures can address to
efficiency affected productivity growth? lift productivity.
• Where is growth heading? What are the potential • Reduced private capital formation since the global
trajectories for medium-term growth given demo- financial crisis in many advanced and emerging
graphic trends and prevailing economic forces, such market economies has also contributed to the growth
as higher debt burdens, geoeconomic fragmentation, decline. Deterioration in firms’ valuations relative
and the emergence of artificial intelligence (AI)? to the cost of capital and rising corporate leverage
What policies could enable a return to the higher are the two most important firm-specific factors
growth rates seen in the two decades preceding contributing to the decline in business investment.
the pandemic? At the macroeconomic level, lackluster growth

66 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Figure 3.2. Five-Year-Ahead Real GDP Forecast by Country: Figure 3.3. Five-Year-Ahead Real GDP Forecast by Regions,
April 2008 versus April 2024 2008, 2019, and 2024
(Percent) (Percent)

10 AEs 45-degree line 10

EMDEs Apr. 2008 WEO Oct. 2019 WEO Apr. 2024 WEO

8 8
IND
April 2024 WEO

6 6
CHN
4 USA 4
FRA IDN

DEU
2 2
BRA
JPN GBR RUS
0 0
0 2 4 6 8 10 World USA and ECA EAP SA LAC MENA SSA
April 2008 WEO Canada

Source: IMF staff calculations. Source: IMF staff calculations.

Note: Bubble size reflects size of the economy using April 2024 GDP in Note: The figure uses GDP in purchasing-power-parity international dollars from
purchasing-power-parity international dollars. Data labels in the figure use the corresponding vintages for aggregation. EAP = East Asia and Pacific; ECA =
International Organization for Standardization (ISO) country codes. AEs = advanced Europe and Central Asia; LAC = Latin America and Caribbean; MENA = Middle
economies; EMDEs = emerging market and developing economies; WEO = World East and North Africa; SA = South Asia; SSA = sub-Saharan Africa; USA = United
Economic Outlook. States; WEO = World Economic Outlook.

performance and uncertainty have inhibited invest- Insights from Medium-Term Forecasts
ment in advanced economies. Five-year-ahead WEO growth projections show
• Demographic pressures weighing on labor supply are a broad-based downturn in growth prospects since
expected to intensify in the medium term in most 2008 that affects nearly 82 percent of economies,
advanced economies and major emerging markets, including the world’s largest (Figure 3.2). Notably,
contributing to lower global growth. By 2030, global the five largest emerging market economies—Brazil,
labor supply growth is projected to be a mere China, India, Indonesia, and Russia—contributed
0.3 percent, less than a third of its average in the approximately 0.8 percentage point of the 1.8 per-
decade before the pandemic. centage point drop in projected global growth.
• Confronted with several structural headwinds, return- The downshift is evident across different regions
ing global growth to its historical average requires and most pronounced for East Asia and the Pacific
both strong policy support and harnessing the poten- (Figure 3.3).
tial of emerging technologies. Based on projected The dimming growth outlook raises two ques-
demographic trends and conservative assumptions tions. First, could it be driven by growing pessimism
about technological progress, global growth in the among forecasters, especially after recent global shocks?
medium term could fall below 3 percent. Return- Tracking the average discrepancy between forecast and
ing to the historical (2000–19) annual growth realized growth shows no evidence of pessimism bias
average of 3.8 percent requires growth-enhancing (Online Annex Figure 3.1.1).2 The subdued prospects
policies and reforms. Their implementation should could in part reflect a correction to previous optimism,
aim to improve allocative efficiency and labor especially since 2012. Second, to what extent does
participation and facilitate cross-border trade and the dimming outlook reflect secular growth trends?
knowledge exchange. These policies and reforms Forecasters typically consider the medium term the
should also enhance innovation capabilities and
maximize the capacity to benefit from technological 2All online annexes are available at www​.imf​.org/​en/​

advances such as AI. Publications/​WEO.

International Monetary Fund | April 2024 67

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

horizon during which economies close the gap between Figure 3.4. Contribution of Components of GDP Growth,
actual and potential output. Indeed, the evidence 1995–2023
suggests that WEO medium-term growth forecasts (Percent)
are generally well aligned with projections of potential Capital Labor TFP Real GDP Real GDP per capita
output growth (Online Annex Figure 3.1.2). Devia-
5 1. World 2. AEs 5
tions have occurred only after crises when forecasters
expected faster growth (relative to potential) to close a 4 4
large output gap.
The decline in global growth forecasts may in 3 3
part reflect progress in living standards and a subse-
2 2
quent slowdown in growth rates. However, when the
historical pace of income convergence across coun- 1 1
tries is considered, the catch-up efforts of emerging
market and developing economies explain only about 0 0
1995– 01–07 08–19 20–23 1995– 01–07 08–19 20–23
a quarter of the projected global growth decline since 2000 2000
2008 (see Box 1.1 of the October 2023 WEO). In
addition, the more accelerated decline in growth 8 3. EMMIEs 4. LIDCs 8
prospects in these economies, compared with that
in advanced economies, poses concerns about future 6 6
convergence. Using various measures, Box 3.2 suggests
that the pace of convergence in regard to income and 4 4
social welfare is slowing or potentially reversing over
the medium term—in stark contrast to prepandemic 2 2
historical trends.
0 0
1995– 01–07 08–19 20–23 1995– 01–07 08–19 20–23
2000 2000
How Did We Get Here?
World growth accelerated from the early 2000s Sources: International Labour Organization; Penn World Table version 10.01;
until the global financial crisis in 2008 and has United Nations, World Population Prospects; and IMF staff calculations.
Note: Growth decomposition sample comprises 140 countries. Contributions of
declined ever since (Figure 3.4), aligned with the capital growth and labor growth reflect output share of respective factor inputs and
dynamics of medium-term projections. This pattern their growth rates. AEs = advanced economies; EMMIEs = emerging market and
middle-income economies; LIDCs = low-income developing countries; TFP = total
has been reflected in both emerging market econo- factor productivity.
mies and low-income countries, mirroring the ebbs
and flows in globalization that have affected capital
flows and productivity. Advanced economies, how-
from 1.3 percent during 1995–2000 to 0.2 percent
ever, have experienced declining growth, beginning
after the pandemic, accounting for half of the GDP
in the early 2000s.3 In per capita terms, GDP growth
growth reduction. Similarly, in emerging market
has followed a similar trend in all country groups,
economies and low-income countries, TFP growth
with a modestly smaller postcrisis decline as popula-
dropped from 2.5 percent and 2 percent, respectively,
tion growth has slowed.
during 2001–07 to just 0.7 percent and nearly zero,
For all country groups, these shifts in growth have
respectively, after the pandemic. In addition, slower
primarily been the result of changes in TFP growth.
capital formation after 2008 for advanced economies
In advanced economies, annual TFP growth fell
and since 2013 for emerging market economies has
also contributed to the global growth slowdown. A
3GDP mismeasurement with expansion of the digital economy
consistent decline in the labor contribution as a result
is often mentioned as a potential explanation for the productivity
slowdown, particularly in the United States. The quantitative of an aging population and a related retreat in labor
relevance of this issue, however, remains an open question. For force participation in major economies have also
instance, Syverson (2017) provides evidence that challenges the played a role.
“mismeasurement hypothesis”; Crouzet and Eberly (2021) estimate
that it may account for a significant share of the decline in TFP and, This section examines each component of output
consequently, GDP growth. growth to understand the drivers behind their trends.

68 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Figure 3.5. Slowdown in the Growth of the Working-Age Figure 3.6. Breakdown of Change in Labor Force
Population, 2008 versus 2021 Participation Rate, 2008–21
(Growth in the working-age population, percent) (Percentage points)

5 AEs 45-degree line 6 Aging effect

Working-age population growth rate in 2021

EMMIEs Within-group change (female)

4 LIDCs 4 Within-group change (male)
Total change
3
2
2 USA IDN
IND 0
GBR
1
FRA
BRA –2
0
JPN
–1 –4
CHN
DEU RUS
–2 –6
–2 –1 0 1 2 3 4 5 Adv. MENA Europe LAC World SSA USA Emerg. CHN
Working-age population growth rate in 2008 Asia and Asia ex.
Pacific CAN CHN
Sources: United Nations, World Population Prospects; and IMF staff calculations.
Note: Working-age population is defined as people ages 15 to 64. Outlier countries Sources: International Labour Organization; and IMF staff calculations.
are excluded to enhance presentation. Including them does not change the pattern. Note: Adv. Asia = advanced Asia; CAN = Canada; CHN = China; Emerg. Asia =
Data labels in the figure use International Organization for Standardization (ISO) emerging Asia; ex. = excluding; LAC = Latin America and the Caribbean; MENA =
country codes. AEs = advanced economies; EMMIEs = emerging market and Middle East and North Africa; SSA = sub-Saharan Africa; USA = United States.
middle-income economies; LIDCs = low-income developing countries.

participate in the labor market. Shift-share analy-

A Demographic Drag on the Labor Supply sis helps tease out some effects of aging and gender
As a country undergoes a demographic transition, disparities in labor force participation on aggregate
with declining fertility rates and an aging population, participation rates (Figure 3.6). First, aggregate labor
the share of its working-age population starts to shrink. force participation rates declined somewhat signifi-
Several large economies (Canada, China, United cantly between 2008 and 2021 in most world regions,
Kingdom, United States) experienced this turning except Advanced Asia and the Pacific, the Middle
point around the time of the global financial crisis East and North Africa, Europe, and Canada. Second,
(Online Annex Figure 3.2.1), in line with a noticeable the drag on participation from aging is visible in all
decline in labor’s contribution to growth (Figure 3.4). advanced economies and China, and to a lesser extent
Since 2008, growth in the working-age popula- in Latin America. Third, advanced economies—except
tion (ages 15–64) has slowed in about 92 percent of the United States—managed to counter this aging
the global economy and has been negative in about effect by significantly increasing their within-group
44 percent (Figure 3.5). The slowdown is visible in labor force participation, mostly through impressive
most advanced and emerging market economies, gains in female participation and higher participation
whereas low-income countries still enjoy a demo- of older workers. The decline in average hours worked
graphic dividend. These demographic shifts have a in Europe (Astinova and others 2024) may have coun-
direct bearing on global labor supply. Countries with tered some of these gains. Last, for emerging market
a current demographic dividend could help support economies and the United States, the decline in male
growth in the global workforce, in which nearly two participation was a drag on aggregate participation.
in every three new entrants over the medium term will Although these trends were evident before 2019, the
come from India and sub-Saharan Africa. The global pandemic shock has exacerbated the drop in partici-
imbalance in labor supply also hints at the importance pation somewhat, especially in emerging markets. The
of migrant workers for advanced economies. initial pandemic shock led to a strong retraction in
As the labor force ages and the share of older participation rates between 2019 and 2020, especially
workers increases, aggregate labor force participation in China and Latin America, with some recovery in
may also suffer, since older workers are less likely to 2021. That noted, participation remained broadly

International Monetary Fund | April 2024 69

 WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

Figure 3.7. Policies and Labor Force Participation by Gender Figure 3.8. Real Business Investment in OECD Countries
and Age (Index, 2008 = 100)
(Change in labor force participation rate, percentage points)
500 Actual
6 Policy impact Contribution from growth slowdown
90 percent confidence interval 400 Pre-GFC trend
5
300
4
200
3
100
2
0
1
–100
2000 04 08 12 16 21
0
Tax wedge
(M: 25 to 54)

Unemployment benefits
(M: 25 to 54)

Secondary education
(F: 25 to 54)

Childcare
(F: 25 to 54)

LM programs
(F: 25 to 54)

Retirement age
(All: 55 to 64)

LM programs
(All: 55 to 64)
Sources: Organisation for Economic Co-operation and Development (OECD); and
IMF staff calculations.
Note: The figure plots the aggregate business investment for the 21 OECD
economies listed in Online Annex 3.2. Actual and predicted real business
investment growth are cumulated from 1999 and indexed at 100 in 2008. Predicted
values for investment growth are obtained by multiplying the estimated investment-
output elasticity reported in Online Annex Table 3.2.3 by output growth. Weaker
economic activity is defined as a deceleration in output growth. Pre-GFC trend is
Sources: International Labour Organization; Organisation for Economic the expected linear path of the business investment index in 2002–08. Shaded area
Co-operation and Development; and IMF staff calculations. denotes the 90 percent confidence interval. GFC = global financial crisis.
Note: The estimated policy impact is due to a change in the policy variable from
the 75th to the 25th percentile within the distribution of policy variation in the
sample, and where the change is aimed at enhancing labor force participation. The estimates suggest that reduced unemployment
The sample comprises 26 advanced economies and 3 emerging market
economies. F = female; LM programs = labor market programs; M = male. benefits and lower labor taxes are associated with
higher participation for men of prime working age.
For women, an expansion in secondary education
lower than in 2019, especially in Latin America, where enrollment has a positive association with future
participation declined about 1.9 percentage points, and participation rates. Similarly, labor market programs
in the United States, where it lost about 1.4 percent- (such as retraining and reskilling) and childcare
age points.4 programs appear to be supportive. For older workers,
Besides cyclical and structural factors, policies can retirement-age reforms and spending on labor market
also improve labor participation rates.5 To understand programs are also associated with higher participation,
how policy variations may have contributed to differ- which is of particular importance since the population
ences across countries, Figure 3.7 shows the estimated share of this group is on the rise.
impacts of selected policy changes on the participation
of different gender-age groups.
Anemic Private Capital Formation
4More recent data for 2022 for a subset of the economies in The second proximate driver of economic growth
the sample reveal upward revisions for participation rates in Chile, is capital formation. In Organisation for Economic
Colombia, India, and Thailand. In addition, more recent esti-
mates for labor force participation in the United States suggest
Co-operation and Development economies, business
some recovery. investment—the bulk of total investment—tumbled
5To explain the potential role of policies, the chapter estimates a
after 2008, and in 2021 it fell by about 40 percent of
country panel regression to investigate how participation rates for
its pre-global-financial-crisis trend (Figure 3.8).
different age and gender groups respond to policies. This exercise
covers only Organisation for Economic Co-operation and Devel- This section starts by examining whether the
opment (OECD) countries, since data on policy variables for slowdown in economic activity since the 2008 global
non-OECD countries are lacking (see Online Annex 3.2 for details). financial crisis has impeded economy-wide business
Given the potential endogeneity of the policies, the results of this
exercise should be interpreted as associational and not neces- investment. It uses “narrative fiscal shocks”—fiscal
sarily causal. policy changes aimed at reducing budget deficits, likely

70 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Figure 3.9. Net Investment Rates in Advanced and Emerging Figure 3.10. Contribution of Firm- and Macro-Level
Market Economies Determinants to Changes in the Investment Rate since 2008
(Percent) (Percentage points)

14 AEs 2
Tobin’s q Leverage
EMMIEs Other firm-level determinants Past GDP growth
Uncertainty Capital inflows
1 Observed change
12

0
10
–1

8
–2

6 –3
2001 05 09 13 17 21 AEs EMMIEs

Sources: Thomson Reuters Worldscope; and IMF staff calculations. Sources: Ahir, Bloom, and Furceri 2022; Thomson Reuters Worldscope; and IMF
Note: The net investment rate is computed as aggregate investment over staff calculations.
aggregate lagged capital stock net of depreciation. See Online Annex 3.2 for Note: The black diamonds represent the average change in investment rates for
details. The numerator is computed by summing firm-level net investment at the AEs and EMMIEs since 2008 compared with the period before 2008. For AEs,
country-year level; the denominator is computed by summing firm-level capital at pre-2008 averages are computed over 2000–08. For EMMIEs, pre-2008 refers to
the country-year level. The figure plots the average ratio for AEs and EMMIEs using 2006–08. Each layer in the bars represents the average change in the
GDP in purchasing power parity in international dollar weights. AEs = advanced corresponding regressor multiplied by its estimated coefficients. Only regressors
economies; EMMIEs = emerging market and middle-income economies. with significant coefficients are included. Changes are aggregated at the country
level using as weights the relative capital share of each firm. Averages for AEs and
EMMIEs are computed using GDP in purchasing power parity in international
dollar weights. AEs = advanced economies; EMMIEs = emerging market and
not responding to economic conditions—as an instru- middle-income economies; Tobin’s q = the ratio of the market value to the book
mental variable to analyze the investment-output rela- value of a firm’s assets.
tionship.6 The results show that for every 1 percentage
point decline in output growth that is not triggered by
a contraction in business investment, there is a corre- 32 advanced economies and 13 emerging markets (see
sponding 2 percentage point decrease in investment Online Annex 3.2 for details). Figure 3.9 plots the net
growth. This estimated output-investment relationship investment rate—defined as investment divided by
is used to calculate the investment shortfall from the lagged capital stock net of depreciation—aggregated
growth slowdown following the global financial crisis. across the sample economies. Importantly, both invest-
Comparing with the precrisis trend, Figure 3.8 suggests ment and capital stock figures account for intangi-
that as of 2021, about half of the shortfall in business bles, which are crucial for understanding investment
investment since 2008 can be linked to weaker eco- dynamics (see Online Annex 3.2). Consistent with
nomic activity. investment trends in Organisation for Economic
This exercise, however, provides only a partial view Co-operation and Development countries (Figure 3.8),
of investment determinants. To gain further insights the figure shows net investment rates in advanced and
into constraints on investment, besides economic activ- emerging market economies declining after 2008.
ity, the chapter explores the characteristics of firms that The chapter uses regression analysis with firm-level
reduced their investment. data to shed light on the most important firm- and
Using firm balance sheet and income statement macro-level factors determining the investment decline
data, the analysis examines publicly listed firms in since 2008 (see Online Annex Table 3.2.5). The findings
align with theoretical expectations: investment rates
6The narrative fiscal shocks are used as instruments for output increase with a firm’s market value relative to its cost of
growth to address endogeneity concerns that result from simultane- capital (“Tobin’s q”), profits, and cash stock but decrease
ous feedback between investment and output (see Online Annex 3.2 with higher corporate leverage and the cost of debt.
for details). They are constructed based on Pescatori and others Figure 3.10 shows that the overall investment rate
(2011) and extended to 2021 for 21 OECD economies. The p-value
of the first-stage F-statistic is below 0.1 percent, indicating that the has declined, on average, by about 2.3 percentage
narrative fiscal shocks are relevant in explaining output growth. points in advanced economies and 2 percentage points

International Monetary Fund | April 2024 71

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

in emerging markets. Of that investment decline, Figure 3.11. Contribution of Allocative Efficiency to Annual
the regression analysis reveals that more than half in TFP Growth, 2000–19
advanced economies and virtually all in emerging (Percentage points)
markets can be explained by the determinants included 2
in the analysis.
Since 2008, Tobin’s q, an indicator of firms’ future 1

productivity and profitability expectations, has 0

decreased by 10 to 30 percent on average, contribut-
–1
ing to the bulk of the explained decline in investment
in both advanced and emerging market economies –2
(Figure 3.10). In emerging markets, the 20 percent –3
average increase in leverage after 2008 is notable as a
–4
factor in the overall fall in investment rates (see Online
Annex Figure 3.2.4). –5
The decline in GDP growth since 2008 helps
–6
explain the investment decline, even after key AEs EMMIEs
firm-level investment determinants are controlled
Sources: Bureau van Dijk Orbis; EU KLEMS database; Organisation for Economic
for. Rising uncertainty after 2008 makes a smaller Co-operation and Development, Trade in Value Added; and IMF staff calculations.
but still significant contribution to the investment Note: Sample comprises 13 goods and 6 services sectors and 20 economies:
AUT, BEL, BGR, CHE, CHN, CZE, DEU, ESP, EST, FRA, ITA, JPN, KOR, POL, PRT,
decline in advanced economies. In emerging markets, ROU, RUS, SVK, SVN, and USA. See Online Annex 3.2 for details. The black lines in
increased capital inflows since 2008 have been positive the bars represent the median, the bars the interquartile range, and the whiskers
for investment. the minimum and maximum values across samples in the group. Country list uses
International Organization for Standardization (ISO) country codes. AEs = advanced
economies; EMMIEs = emerging market and middle-income economies; TFP =
total factor productivity.
Productivity and the Role of Resource Misallocation
TFP growth has slowed over the past two to three
decades. Previous studies suggest several contributors to during 2000–19 in most countries in a sample of
this trend, particularly affecting within-firm productiv- 15 advanced and 5 emerging market economies
ity. These include waning gains from information and (Figure 3.11).7 The median country in the sample
communication technology (Fernald 2015); declining experienced an average annual drag on TFP growth of
business dynamism (Decker and others 2016; Akcigit about 0.9 percentage point from declining allocative
and Ates 2021); tighter credit conditions, limiting new efficiency. For the median advanced economy, this
technology investments (Adler and others 2017; Duval, drag was 0.5 percentage point. Given that the median
Hong, and Timmer 2020); and a slower expansion of advanced economy saw TFP growth of only 0.5 per-
cross-border capital flows and trade since 2008. cent during this period, this suggests that increased
This section documents the contribution of rising misallocation of capital and labor may have halved its
misallocation of capital and labor to the decline in TFP growth. A notable exception is the United States,
TFP growth and draws lessons for medium-term
growth. So-called allocative efficiency measures the
extent to which capital and labor are allocated to 7Allocative
efficiency measures, approximately, the extent to which
an economy’s most productive firms (see Box 3.1). value added per factor input varies across firms in a given sector. If
the variation is large, there are potentially large gains from reallocat-
A decline in allocative efficiency, whereby resources ing capital and labor among firms, and allocative efficiency is low; if
become more concentrated in relatively unproductive the variation is small, allocative efficiency is high. For each sample
firms over a period of time, can reduce TFP growth; economy, allocative efficiency is computed at the level of 19 broad
sectors, using data from Orbis. The data cover the whole economy,
an improvement in allocative efficiency, as resources including both goods- and service-producing sectors, but the analysis
move toward more productive firms, will, however, excludes predominantly nonmarket sectors (such as health care, edu-
boost TFP growth. cation, and public administration). Sector-level allocative efficiency is
then aggregated using sectors’ shares in whole-economy value added.
The approach used here, pioneered by Hsieh and
See Online Annex 3.2 for details. See G20 (2021) for a discussion
Klenow (2009) and refined by Bils, Klenow, and of the possible impact of the COVID-19 pandemic on allocative
Ruane (2021), finds that allocative efficiency declined efficiency in the post-2019 period.

72 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Figure 3.12. Contribution of Allocative Efficiency to Annual Figure 3.13. TFP Loss from Misallocation, by Sector Type,
TFP Growth, 2000–19 2019
(Percentage points, decomposed) (Percent)

1 100

0
80

–1
60
–2
40
–3
Sector shares
Within sectors 20
–4 Change in allocative efficiency

–5 0
Sample USA AEs ex. USA CHN EMMIEs ex. Goods Services
CHN
Sources: Bureau van Dijk Orbis; EU KLEMS database; Organisation for Economic
Sources: Bureau van Dijk Orbis; EU KLEMS database; Organisation for Economic Co-operation and Development, Trade in Value Added; and IMF staff calculations.
Co-operation and Development, Trade in Value Added; and IMF staff calculations. Note: The figure shows the distribution of calculated total factor productivity (TFP)
Note: Sample comprises 13 goods and 6 services sectors and 20 economies: losses relative to a benchmark of no misallocation (see Online Annex 3.2) for all
AUT, BEL, BGR, CHE, CHN, CZE, DEU, ESP, EST, FRA, ITA, JPN, KOR, POL, PRT, sample countries and sectors in 2019, grouped by sector type. The black lines in
ROU, RUS, SVK, SVN, and USA. The darker shade of colors denotes “within the bars represent the median, the bars the interquartile range, and the whiskers
sectors,” while the lighter shade of colors denotes “sector shares.” Country list the minimum and maximum values across samples in the group. Sample
uses International Organization for Standardization (ISO) country codes. AEs ex. comprises 13 goods and 6 services sectors and 20 economies: AUT, BEL, BGR,
USA = advanced economies excluding United States; CHN = China; EMMIEs ex. CHE, CHN, CZE, DEU, ESP, EST, FRA, ITA, JPN, KOR, POL, PRT, ROU, RUS, SVK,
CHN = emerging market and middle-income economies excluding China; TFP = SVN, and USA. Country list uses International Organization for Standardization
total factor productivity. (ISO) country codes.

where improvements in allocative efficiency helped with regard to productivity and inputs in services.8 As
boost annual TFP growth by 0.8 percentage point over a result, an economy—such as China’s—experiencing
the period. structural transformation from goods to services will
What explains the decline in allocative efficiency register a decline in overall allocative efficiency.
across a large group of economies? The observed A large part of the observed decline in allocative effi-
drag on TFP growth could reflect either decreased ciency within sectors can be traced to uneven firm pro-
efficiency within sectors or a growing share of ductivity growth during some of the 2000–19 period.
already-misallocated sectors in an economy. Analy- As Figure 3.14 shows, the dispersion of firms’ real pro-
sis for the 20 economies shows that changing sector ductivity in the 20 sample economies rose significantly
shares in GDP contributed only about 30 percent leading up to the global financial crisis and, despite
of the annual drag on TFP, with the rest attribut- some subsequent reversion, remains elevated. This
able to within-sector developments (Figure 3.12). aligns with the decline in allocative efficiency, most of
The shift in sectoral GDP shares is an important which also occurred in the first decade of the 2000s.
factor for just a few economies—most significantly
for China, for which it contributes 60 percent of 8Several studies have documented this pattern, using firm-level

the allocative-efficiency impact on TFP growth. The data for a range of countries, such as Hsieh and Klenow (2009),
reason the sectoral composition of the economy affects Busso, Fazio, and Algazi (2012), Devries and others (2011), Dias,
Marques, and Richmond (2016), and Chapter 2 of the April 2017
aggregate allocative efficiency is that sectors differ
Fiscal Monitor. The literature has tended to attribute these patterns
systematically in the measured extent of their misal- to differences in market structure and firm dynamics in goods and
location. Specifically, Figure 3.13 shows that service service sectors. Online Annex 3.2 uses a method proposed by Bils,
sectors display more inefficiency than goods-producing Klenow, and Ruane (2021) to show that there is little evidence that
additive measurement error is more prevalent in service sectors than
sectors. This may reflect structural differences between in goods sectors, but this still leaves room for other types of mea-
goods and service sectors or measurement challenges surement errors to explain some of the difference.

International Monetary Fund | April 2024 73

 WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

Figure 3.14. Dispersion of Firm Productivity, 2000–19 Figure 3.15. Countries’ Structural Allocative Efficiency and
(Index, 2000 = 100, weighted average) Policies
(Log points, USA = 0)
125
AEs EMMIEs Fitted line
120
0.5 1. OECD Product 2. IMF-SRD Trade 0.5
115 Market Barriers Liberalization

Structural allocative efficiency

Structural allocative efficiency

USA USA
0.0 0.0
110

105 –0.5 –0.5

CHN CHN
100
–1.0 –1.0
95
–1.5 –1.5
90 1.0 1.5 2.0 2.5 3.0 0.6 0.7 0.8 0.9 1.0
2000 05 10 15 19 Index, 0–1
Index, 0–3

Sources: Bureau van Dijk Orbis; EU KLEMS database; Organisation for Economic 0.5 3. IMF-SRD Financial 4. IMF-SRD Labor 0.5
Co-operation and Development, Trade in Value Added; and IMF staff calculations. Liberalization Market Liberalization
Structural allocative efficiency

Structural allocative efficiency

Note: Following Bils, Klenow, and Ruane (2021), productivity dispersion is
computed at the sector level as the ratio of the power mean to the geometric USA USA
0.0 0.0
mean of firm output-based total factor productivity (TFPQ)—a measure of the
technical efficiency of a plant. Productivity dispersion is aggregated to the country
level using sector GDP shares. Line shows the three-year moving average, –0.5 –0.5
aggregating across sample economies using GDP in purchasing power parity in CHN
international dollar weights. Value for the year 2000 normalized to 100. Sample
comprises 20 economies: AUT, BEL, BGR, CHE, CHN, CZE, DEU, ESP, EST, FRA, –1.0 CHN –1.0
ITA, JPN, KOR, POL, PRT, ROU, RUS, SVK, SVN, and USA. Country list uses
International Organization for Standardization (ISO) country codes.
–1.5 –1.5
0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0
A widening of the distribution of firms’ real pro- Index, 0–1 Index, 0–1
ductivity has implications for allocative efficiency.
Ideally, firms with rapidly increasing real productivity Sources: Organisation for Economic Co-operation and Development (OECD); and
IMF staff calculations.
should attract capital and labor from those growing Note: The country-specific structural component of allocative efficiency is obtained
more slowly, with marginal revenue products kept as a country fixed effect from the dynamic regression described in Online
Annex 3.2. Sample comprises 20 economies: AUT, BEL, BGR, CHE, CHN, CZE,
equalized. However, firm-level evidence points to DEU, ESP, EST, FRA, ITA, JPN, KOR, POL, PRT, ROU, RUS, SVK, SVN, and USA.
frictions that slow this adjustment process (see Online Country list and data labels in the figure use International Organization for
Standardization (ISO) country codes. AEs = advanced economies; EMMIEs =
Annex Table 3.2.7). This leads to an initial decline in emerging market and middle-income economies; IMF-SRD = IMF Structural
allocative efficiency, as faster-growing firms operate Reform Database.
with less capital and labor than optimal. Consistently,
sector-level evidence shows that a rise in a sector’s
dispersion of real firm productivity is accompanied by structural transformation in some countries, will likely
a decline in its allocative efficiency. continue to affect medium-term TFP growth.
However, this phenomenon is transitory. As time The analysis so far implies that the extent of an econo-
passes, firms that have improved productivity faster my’s overall misallocation has two components at any one
than the rest can scale up their capital and labor input, time: a transitory component that reflects an incomplete
and allocative efficiency once again improves. Yet this adjustment by firms to recent shocks and a longer-lasting,
recovery is slow; firm and sector data suggest that it structural component that reflects the efficiency of
takes 9–11 years for allocative efficiency to return markets and quality of institutions that govern them.
halfway to its long-term fundamental level, which Evidence from firm-level analysis suggests that, for the
is shaped by sector characteristics and a country’s economies analyzed, about one-third of measured misallo-
economic and institutional environment (see Online cation is attributable to transitory factors, and two-thirds
Annex Table 3.2.8). Consequently, evidence from has structural roots (see Online Annex 3.2).
sector-level analysis shows that recent shifts in the Figure 3.15 shows wide cross-country variation in
firm productivity distribution, along with ongoing one measure of structural allocative efficiency (along

74 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

the vertical axes and based on the analysis in Online Figure 3.16. Medium-Term Growth Projections of Potential
Annex 3.2), which rises with market entry and com- Employment
petition, trade openness, financial access, and labor (Percent)
market flexibility. While some of these indicators of 3 Population (15+)
market efficiency and barriers broadly improved during Labor force participation rate
the 2000–19 period (notably, trade and financial Employment
2
liberalization), others worsened for some countries in
the sample, with no systematic evidence that changes
in structural policies are behind the observed decline in 1

allocative efficiency over the past two decades.

However, the large cross-country differences in 0
structural allocative efficiency suggest that there is
potential to raise TFP growth through reforms. Analy- –1
sis of the 20 sample economies shows that if countries
whose allocative efficiency is currently lower than –2
that of the United States were to reduce their gaps in LICs EMMIEs USA World AEs EU CHN
ex. CHN ex. USA
structural policies by 15 percent over 10 years, it could and EU
boost medium-term TFP growth by 0.7 percentage
point. While historical instances of such significant Sources: International Labour Organization (ILO); United Nations, World Population
Prospects; and IMF staff calculations.
policy catch-up are not common, they are not unprec- Note: Sample comprises 140 countries. Estimation for labor force participation rate
edented, representing an ambitious yet achievable is based on a cohort model (Online Annex 3.3) using data from ILO for 83 countries.
policy objective. The remaining 57 countries follow the 2014–19 average growth rate in the
participation rates. AEs = advanced economies; CHN = China; EMMIEs = emerging
Improving market efficiency may also make it easier market and middle-income economies; EU = European Union; ex. = excluding;
for firms to adapt to future shocks. Firm data provide LICs = low-income countries; USA = United States.
some evidence that the US avoided an overall decline in
allocative efficiency during the 2000–19 period because tural impacts on labor supply. These, along with United
resources relocated across firms faster as firms’ produc- Nations demographic projections, provide estimates of
tivity dispersion increased. This led to a faster reversal potential employment growth, with stable employment
of the transitory rise in misallocation that has contin- rates assumed. Capital growth projections merge WEO
ued to weigh on TFP for most other sample economies. public investment forecasts with this chapter’s estimates
of the medium-term private investment rate. Finally,
TFP growth is projected by assuming that sectoral
Where Is Growth Heading? allocative efficiency is moving gradually toward its
This chapter’s focus so far has been on analyz- estimated long-term level and reaching its half-life in
ing historical trend growth and the factors behind the medium term, whereas efficient TFP growth—net of
its decline. New tailwinds and headwinds could yet misallocation—follows the historical trend.
further affect growth trajectories. This section shifts • By 2030, the annual contribution of labor supply to
the focus to a forward-looking question: What are global GDP growth is expected to decrease to 0.2 per-
the likely medium-term growth trajectories, and can centage point, only a quarter of its 2000–19 average
annual global growth return to the 3.8 percent average contribution. This reflects a modest 0.3 percent
for 2000–19? projected growth of potential labor supply in 2030
(Figure 3.16). The slowdown reflects falling partici-
pation rates, which dampen the effect of population
Baseline Scenario growth on labor supply. However, trends in labor
This section assesses the prospects of labor, capital, supply vary widely by region. Low-income coun-
and TFP in the medium term, defined as the year tries are expected to experience robust 2.1 percent
2030, drawing on analyses in earlier sections (projection growth in labor supply, highlighting the need for
methods are detailed in Online Annex 3.3). Specifically, job creation to translate this supply growth into
labor force participation forecasts use a cohort-based employment. Meanwhile, labor supply in emerging
approach, considering life-cycle, generational, and struc- market economies, excluding China, will grow by

International Monetary Fund | April 2024 75

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

0.9 percent, and in the US by 0.5 percent, whereas Figure 3.17. Impact of Various Factors on Global
a sharp reduction in participation will cause labor Medium-Term Growth
supply to contract by 0.6 percent in China and by (Relative to the baseline, percentage points)
0.5 percent in the EU. 2.0 Single level
• Capital’s contribution to growth is expected to be High-low range
1.7 percentage points, compared with the 2000–19 1.5
average contribution of 2.1 percentage points. Con-
tinued high public debt will likely constrain future 1.0
public investment in emerging market and devel-
oping economies, which accounts for 30 percent of 0.5
these countries’ overall capital. Advanced economies
0.0
are expected to see a modest increase in public
investment, but its growth impact will be mini-
–0.5
mal given its small share in overall investment. In
addition, private investment rates are expected to –1.0
remain low in both country groups, owing to sub-

Policies boosting
LFPR

Migration boost
to AEs’ labor supply

Structural reforms
reducing misallocation

Policies improving
allocation of talent

AI adoption

Public debt
overhang

Fragmentation
dued economic prospects and the anticipated lower
employment and TFP growth.
• The TFP growth contribution is expected to decline to
0.9 percentage point by 2030, down from the 2000–19
average of 1.0 percentage point. The ongoing decrease
in allocative efficiency is expected to slow TFP growth Source: IMF staff calculations.
to a lesser degree. Meanwhile, the growth in efficient Note: The estimated impact on medium-term growth is presented relative to the
baseline projection for each scenario described in the labels on the horizontal axis.
TFP, which reflects the rate of technological progress, See Online Annex 3.3. The scenarios include policy interventions—aiming at
is expected to slow in the baseline scenario, following increasing labor force participation, supporting AEs’ labor supply through
migration, reducing misallocation, and improving talent allocation in emerging
its long-term trend. Factors such as the increasing market and developing economies—and scenarios in which artificial intelligence
difficulty of generating new ideas (Bloom and others is widely adopted, there is a persistent public debt overhang, and geopolitical
2020), slower growth of research employment (Jones blocs are emerging (“fragmentation”). AEs = advanced economies; AI = artificial
intelligence; LFPR = labor force participation rate.
2023), a plateau in educational attainment, and the
slower catch-up process are expected to play a role.
The net effect is a decline in the TFP growth rate These scenarios assess the effects of policy changes
by 0.1 percentage point from its two-decade average related to labor supply and resource allocation and of
prior to the pandemic. However, major technologi- economic tailwinds and headwinds—positive impacts
cal advances, particularly in AI, could increase TFP of AI and negative effects of public debt overhang
growth substantially. and geoeconomic fragmentation. To gauge the feasi-
bility of the policy scenarios, large and ambitious—
When the contributions of the three factors are but not unprecedented—policy shifts are considered.
summed, the world’s growth rate is projected at Overall, the medium-term growth effects range
2.8 percent in 2030 under the baseline scenario. from 1.2 percentage points above to 0.8 percentage
This suggests that global growth could fall even point below the baseline (Figure 3.17). Larger effects
more, below the current WEO medium-term forecast are possible if these scenarios occur simultaneously.
(see Chapter 1). This would represent a significant However, given high uncertainty surrounding these
slowdown relative to the historical (2000–19) annual estimates, the figures should be viewed as indica-
average of 3.8 percent. tive of the potential impacts (see Online Annex 3.3
for details).
• Policies to increase labor force participation: This
Alternative Scenarios scenario assumes that countries increase their labor
What factors could elevate growth or pose emerg- force participation rates by 3.2 percentage points,
ing risks? This section compares various scenarios the median increment in participation if all countries
against the baseline medium-term growth projection. converged to the best policies. This could increase

76 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

labor supply growth by about 0.3 percentage point, outcomes in three scenarios—one scenario in which
contributing 16 basis points to global growth. debt continues to increase with stable public defi-
• A migration boost to labor supply in advanced econ- cits and two debt-stabilization scenarios in which
omies: Migrant workers have supported growth increased interest payments are offset either by reduc-
in advanced economies by filling labor gaps. This ing transfers or public investment. The overall impact
scenario assumes higher flows, along with enhanced is considered moderate because the scenario does not
labor market integration for migrant workers, that assume extensive fiscal consolidation aimed at signif-
translates into an increase in labor supply equivalent icant debt reduction or additional channels through
to 1 percent of advanced economies’ projected labor which public debt could affect growth (Pattillo, Poir-
force in 2030. The resulting increase in labor supply son, and Ricci 2004; Woo and Kumar 2015).
could add 20 basis points to global growth. • Geoeconomic fragmentation: The emergence of geo-
• Structural reforms for improving allocative efficiency: economic blocs leading to international trade and
Building on the previous section, this scenario assumes foreign direct investment fragmentation could reduce
that countries close 15 percent of their policy gap with capital and knowledge flows significantly and suppress
the United States in areas such as product and labor growth (Chapter 3 of the October 2023 Regional
market policies, trade openness, and financial deepen- Economic Outlook: Asia and Pacific). The April 2023
ing over the medium term. These structural reforms WEO provides reasonable scenarios analyzing the
are expected to greatly reduce the drag from misal- effects of heightened trade barriers. These vary from
location and enhance TFP growth by 0.7 percentage limited cases in which a “US bloc” and a “China
point, which, in turn, could stimulate investment and bloc” engage in some “friend-shoring,” reducing
add 1.2 percentage points to global growth. growth by 10 basis points, to a more extensive
• Improved talent allocation in emerging market and scenario in which all regions reshore some trade,
developing economies: Although gaps in occupation potentially lowering medium-term growth by 80 basis
and earnings between men and women have been points. A greater loss could result from a reduction
narrowing in advanced economies, they remain in trade-associated knowledge spillovers (Ahn and
significant elsewhere. Closing these gaps could lead others, forthcoming) and productivity loss, but it is
to substantial productivity gains, especially if jobs not accounted for in this simulation.
are filled based on innate talent and comparative
advantage, not skewed by social norms, barriers, or The scenario impacts underscore a clear message:
discrimination (Berg and others 2018; Hsieh and regaining historical growth will demand substantial
others 2019; Jayachandran 2021). Should talent policy efforts and, possibly, harvesting net positive ben-
allocations in emerging market and developing efits from AI. Structural reforms to resolve misalloca-
economies follow the trend in the United States over tion are key to restoring growth to historical averages.
past decades, global growth could be boosted by a
quarter of a percentage point.
• AI technologies: AI technologies stand at the brink Conclusions and Policy Recommendations
of transforming many aspects of the world econ- The chapter’s analysis suggests that the global
omy (Cazzaniga and others 2024). Their impact on economy’s declining actual growth and waning growth
economic growth is highly uncertain but potentially expectations largely reflect persistent headwinds. A
substantial. Generally, AI’s enhancement of labor significant slowdown in TFP has emerged as a key fac-
productivity is expected to outweigh its negative tor, with that slowdown driven by increased resource
effects on labor demand. Depending on how widely misallocation and slower growth in efficient TFP. A
it is adopted and whether it replaces or augments shrinking working-age population in major economies,
workers, the estimated global growth impact varies coupled with lackluster business investment, has also
from 10 to 80 basis points in the medium term (see contributed. For the most part, the implications of
Box 3.3 for more details). the analysis here are sobering for medium-term global
• Legacy of high public debt: Persistent elevated public growth prospects. Absent timely policy interventions
debt raises global economic growth concerns, poten- and a boost from emerging technologies, global growth
tially reducing medium-term growth by an estimated is likely to remain well below its prepandemic histori-
5 to 15 basis points. The projection simulates growth cal average in the medium term.

International Monetary Fund | April 2024 77

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

How could policies help elevate growth? The chap- Investment in human capital, especially in
ter’s findings suggest that interventions should focus on low-income developing countries, is essential to
reforms that promote market competition, trade open- leverage their demographic dividend. In regard to
ness, financial accessibility, and labor market flexibility. capital formation, since higher corporate leverage
These could significantly boost TFP growth by alleviat- has held back business investment in emerging
ing institutional and financial barriers that impede the market economies, reforming mechanisms for
efficient allocation of capital and labor across firms. restructuring and insolvency and eliminating debt
Such reforms offer substantial gains for growth and can bias in corporate tax policies can also help support
be complemented by governance and external sector medium-term growth (Chapter 2 of the April 2022
reforms (Budina and others 2023). Industrial policies WEO). To lessen the negative growth impact from
targeted to specific sectors, if poorly designed, may increased geoeconomic fragmentation, it is import-
impede resource allocation to more productive firms or ant to steer clear of damaging unilateral trade and
sectors (see the April 2024 Fiscal Monitor on industry industrial policies.
policy for innovation). The global medium-term prospects are not all doom
At the same time, policies designed to facilitate the and gloom. Resilience amid various shocks (Chapter 1)
flow and integration of migrant workers, alongside and the emerging promise of technologies such as AI
measures to boost labor force participation among older could prove transformative for medium-term global
workers in advanced economies—through retirement growth. To fully harness this potential, countries must
reforms and labor market programs—could mitigate strengthen their regulatory frameworks, including
the increasing demographic pressures on labor supply. intellectual property protection, and revisit redistribu-
Encouraging the participation of women in emerging tive and other adjustment programs to ensure that the
market economies, by expanding education enrollment benefits from AI are shared fairly and widely (Cazza-
and childcare support, could unlock their untapped niga and others 2024). Looking beyond the medium
potential. These efforts should be complemented by term, policies geared toward promoting innovation
policies that reduce social barriers and gender discrimi- play a crucial role in defining the path of future
nation to ensure talent is optimally allocated across jobs. global growth.

78 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Box 3.1. Allocative Efficiency: Concept, Examples, and Measurement

Not only is total factor productivity (TFP) growth high in one firm and low in another, more economic
driven by well-known factors such as technological value would be created by moving resources from the
innovation and novel business practices that enhance second firm to the first. This approach tells us that an
within-firm productivity, it is also propelled by how economy’s allocative efficiency is improving if marginal
well the allocation of capital and labor across firms revenue productivity across firms is becoming more
reflects their relative productivity—known as “alloc- similar and that it is worsening if it is becoming more
ative efficiency.” Consider an example of two firms, dispersed.1
one with high and one with low productivity. If too Achieving lasting improvements in allocative effi-
much capital and labor are tied up in the relatively ciency requires tackling the frictions that slow firms’
unproductive firm, average productivity will be low—a ability to change their scale of operations as needed or
case of poor allocative efficiency. TFP would rise if that permanently favor or penalize some firms irrespec-
capital and labor moved to the more productive firm, tive of their productivity. Many studies have identified
correcting the initial misallocation. the structural sources of these frictions. These include
A variety of frictions can cause capital and labor size-dependent tax, labor, and social insurance policies
to be allocated to the “wrong” firms. Some frictions (Levy 2018; Ulyssea 2018); informality and corrup-
may do so only temporarily. In the two-firm exam- tion (Misch and Saborowski 2018); weak property
ple, the productive firm may be looking to expand, rights (Adamopoulos and Restuccia 2020); regional
but its search for new workers may take time. In this barriers (Tombe and Zhu 2019); restrictive trade
case, allocative efficiency may be low for a while but policies (Khandelwal, Schott, and Wei 2013; Edmond,
will rise as the productive firm gradually attracts new Midrigan, and Xu 2015); uneven firm markups (Peters
employees from its less-productive competitor. How- 2020); and financial frictions (Song, Storesletten,
ever, other frictions may weigh on allocative efficiency and Zilibotti 2011; Midrigan and Xu 2014; David,
more permanently. For example, the unproductive Hopenhayn, and Venkateswaran 2016; Gopinath and
firm may be politically connected and receiving sub- others 2017; Libert 2017). Several country case studies
sidies or tax breaks that allow it to operate on a larger have highlighted specific policies that successfully
scale than its profits merit. reduce misallocation, such as removing barriers to
Measuring the extent of allocative (in)efficiency international trade (Ha and Kiyota 2016) and reforms
in practice is challenging. One influential approach, aimed at correcting distortions in credit access (Chen
developed by Hsieh and Klenow (2009) and used and Irarrazabal 2015).
throughout this chapter, measures it indirectly by
1This is related to, but distinct from, an earlier measure of
comparing the marginal revenue product of capital
and labor across firms—that is, the additional revenue allocative efficiency developed by Olley and Pakes (1996). Oper-
ationalizing the latter requires information on real productivity
that one more unit of capital or labor could earn in
(quantity total factor productivity) at the firm level, which is dif-
any given firm. If marginal revenue productivity is ficult to measure for a large sample of countries and firms. The
approach of Hsieh and Klenow (2009) requires only information
The authors of this box are Nan Li and Robert Zymek. on relative revenue productivity, which is easier to obtain.

International Monetary Fund | April 2024 79

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

Box 3.2. Distributional Implications of Medium-Term Growth Prospects

The medium-term growth slowdown could affect Figure 3.2.2. Global Inequality, 1995–2028
global income inequality and convergence between (Gini points, 0 = perfect equality; times)
countries. A slower growth environment makes it
challenging for poorer countries to catch up with 0.75 Gini 65
Forecast
those that are richer. Slower GDP growth can also Top 10/bottom 50 (right scale)
lead to higher inequality, reducing average welfare. 0.74 60
This box examines the implications in three areas:
between-country convergence, global inequality, and 0.73 55
welfare convergence. Between-country convergence
has been sustained since the global financial crisis. 0.72 50
One way to measure it is to compare countries’ initial
GDP with their subsequent growth. When this rate 0.71 45
is negative, countries with lower levels of income are
growing faster than those with higher levels, implying 0.70 40
convergence. Cross-country convergence took place
during 2008–19 (Figure 3.2.1) and was fastest during 0.69 35
1995 2000 05 10 15 20 25 28
2008–12. However, the rate turned positive after
Sources: World Inequality Database; and IMF staff
calculations.
Note: On the left scale, the Gini index calculates how the
global income distribution deviates from a perfectly equal
Figure 3.2.1. GDP Convergence between distribution. Income is measured before taxes. “Top 10/
Countries, 2000–28 bottom 50” compares the average income of the top 10 and
(Rate at which gap to frontier is closed, negative = bottom 50 of the global income distribution. On the right
scale, “times” refers to the number of times the average
convergence) income of the top 10 of the income distribution is larger than
the average income of the bottom 50 of the distribution. For
1.0 95 percent confidence interval example, a value of 40 on the right scale means the average
Forecast
Historical WEO projections income of the top 10 is 40 times larger than that of the
With AI effects bottom 50.
0.5

0.0 the pandemic. Current projections point to no conver-

gence over the medium term.
–0.5
The previous computation does not consider how
the gains from convergence are distributed within a
country, only country averages (“between-country”
–1.0 inequality). Milanovic (2002) and Chancel and Piketty
(2021) estimate measures of global income distribution
–1.5 and inequality, the comparison of the income position
2000 04 08 12 16 20 24 28 of a group of people in one country with those of
other groups in the world. These measures show that
Source: IMF staff calculations.
Note: The convergence rate for year t corresponds to the βt although inequality has decreased since the mid-2000s,
coefficient in the following regression: Δlog(GDPpcit) = αt + the pandemic reversed some of the gains (Figure 3.2.2;
βtlog(GDPpcit − 5) + εit, in which Δlog(GDPpcit ) is average World Bank 2022). While between-country conver-
year-over-year GDP per capita growth in the five-year period
between t and t – 5 and log(GDPpcit − 5) is GDP per capita at gence has driven the reduction in global inequality in
the beginning of the period. See Box 3.3 for effects of the past two decades, most of this inequality now stems
artificial intelligence (AI effects) on growth. AI = artificial from differences within countries.1
intelligence; WEO = World Economic Outlook.

1Sovereign governments usually engage in policies that affect

within-country inequality. The analysis presented here uses pre-

The authors of this box are Gabriela Cugat and Carlos tax data to focus mainly on changes in inequality derived from
van Hombeeck. economic trends before government intervention.

80 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Box 3.2 (continued)

To assess the impact of the medium-term outlook, a Figure 3.2.3. GDP Growth and Welfare
projection for global inequality is created by combin- Drivers before and after the COVID-19
ing within-country and between-country inequality Pandemic
projections derived from the World Economic Outlook (Percentage points)
(WEO).2 Depending on the measure analyzed, there
is either no or only a modest expected recoupment in 0.3 Consumption Inequality
Leisure Life expectancy
the medium term (Figure 3.2.2). Small within-country 0.0
inequality improvements are not sufficient to offset
–0.3
the expected slowdown in between-country inequality
convergence. –0.6
The results use GDP as a proxy for welfare, but this –0.9
association could be flawed (Coyle 2017), since it does
–1.2
not include unpaid household work or the environmen-
tal cost of economic growth, for example. Jones and –1.5
Klenow (2016) propose a welfare measure, based on –1.8
lifetime expected utility, that complements consump-
–2.1
tion (highly correlated with GDP) with life expectancy,
leisure, and (less) inequality. Welfare growth histori- –2.4
GDP Welfare GDP Welfare GDP Welfare
cally has exceeded GDP growth, driven mostly by life growth growth growth growth growth growth
expectancy improvements (see Box 1.2 of the October AEs EMEs LIDCs
2020 WEO). Across the board, both GDP and welfare
growth are predicted to fall in the postpandemic period Sources: Penn World Table version 10.01; United Nations
Population Division; World Bank, World Development
(Figure 3.2.3). Welfare growth is expected to deteriorate Indicators; and IMF staff calculations.
more than GDP growth, driven by stalled dimensions Note: This figure shows the difference in average annualized
such as life expectancy and within-country inequality, GDP growth and welfare growth between 2010–19 and
2024–28. The components of the difference in welfare
leading to welfare divergence between countries. growth are listed in the legend. AEs = advanced economies;
The growth slowdown has grim implications for the EMEs = emerging market economies; LIDCs = low-income
distribution of income between countries, of global developing countries.
income, or of a more general welfare measure. Based
on results from Box 3.3, the expected skewed effect
of artificial intelligence on growth would increase in Figure 3.2.1). Inasmuch as other factors, such as
between-country divergence (the “with AI effects” line geoeconomic fragmentation, worsen the distribution
of income between countries, they will likely worsen
2Within-country inequality projections are based on how
global inequality and the distribution of welfare, unless
GDP growth is distributed within a country. See Cugat, Li, and
they significantly improve income distribution within
van Hombeeck (2024) for more details on how the distribution countries and other dimensions of welfare, such as
of growth within countries is estimated. life expectancy.

International Monetary Fund | April 2024 81

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

Box 3.3. The Potential Impact of Artificial Intelligence on Global Productivity and Labor Markets
Artificial intelligence (AI) stands at the forefront Figure 3.3.1. Employment Shares by AI
of a transformative wave, often equated with a new Exposure and Complementarity
industrial revolution, with the potential to reshape the (Percent of employment)
global economy. While its profound and far-reaching Low complementarity
economic and social consequences are not yet fully High exposure, low complementarity
understood, AI’s impact on the global economy High exposure, high complementarity
exhibits a clear dichotomy. On one hand, AI holds
100
the promise of enhancing productivity. On the other,
it poses a formidable challenge, with the potential to
replace humans in certain jobs and fundamentally alter 80
the nature of others.
Building on AI’s potential diverse impacts, IMF 60
staff have advanced a nuanced framework to assess
AI’s influence on productivity and the labor market.
This approach, based on the concept of AI “exposure” 40
(Felten, Raj, and Seamans 2021, 2023), is extended by
the AI complementarity concept (Pizzinelli and others 20
2023), which delivers new insights into the likelihood
of jobs’ either benefiting from AI or being at risk.
0
There is significant disparity in AI exposure between World AEs EMs LICs
country groups—approximately 60 percent of jobs
in advanced economies are susceptible to changes as Sources: Cazzaniga and others 2024; International Labour
Organization; and IMF staff calculations.
a result of AI, compared with 40 percent in emerg- Note: Share of employment within each country group is
ing market economies and 26 percent in low-income calculated as the working-age-population-weighted
countries (Figure 3.3.1; Cazzaniga and others 2024). average. AEs = advanced economies; AI = artificial
intelligence; EMs = emerging markets; LICs = low-income
In advanced economies, AI is expected to enhance countries; World = all countries in the sample.
productivity in half of these exposed jobs, signaling a
positive impact. For the other half, AI integration could
automate tasks, potentially reducing labor demand and
wages and even leading to job obsolescence. In contrast, dom, a country highly exposed to AI adoption and for
emerging market and developing economies are less which data on households’ asset holdings are available.
likely to experience immediate disruption but may also The impact of AI on productivity is analyzed
see fewer benefits from AI. Many lack the necessary through two scenarios. In the first (high comple-
infrastructure and skilled workforce to effectively lever- mentarity), AI significantly enhances roles with
age AI technology, raising concerns that, over time, AI strong complementarity. The second scenario
could exacerbate inequality across countries. (high complementarity and high productivity) expands
A model-based analysis gauges AI’s potential impact this complementarity by having AI also boost overall
on productivity. In this model, AI affects productivity productivity, enhancing the high-complementarity role
through three critical channels: labor displacement, AI (see Rockall, Pizzinelli, and Tavares 2024 on the mod-
complementarity with skills, and productivity gains. eling analysis and Cazzaniga and others 2024 for more
First, AI adoption may shift tasks from humans to information about the distributional implications.)
AI-driven systems, enhancing the efficiency of task In the first scenario, AI use leads output to increase
completion. Second, AI integration could benefit by almost 10 percent as the UK economy adjusts to
tasks that are highly complementary with AI. Third, the new steady state through a combination of capital
AI adoption may lead to broad-based productivity deepening and a small increase in total factor pro-
gains, boosting investment and increasing overall labor ductivity (Figure 3.3.2). In the second scenario, when
demand. The model is calibrated to the United King- the productivity impact is also considered, output
expands by 16 percent and total factor productivity
increases by almost 4 percent. These gains take place
The author of this box is Marina M. Tavares. primarily in the first decade of transition. Incomes

82 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Box 3.3 (continued)

Figure 3.3.2. Impact of AI on TFP and Output for all workers increase, ranging from 2 percent
in the United Kingdom for low-income workers to almost 14 percent for
(Percent) high-income workers, leading to higher income
inequality.
16
TFP Productivity gains from AI are expected to range
Output from 0.9 to 1.5 percent a year, thanks to the United
14
Kingdom’s robust digital infrastructure, skilled
12 labor force, innovation ecosystem, and regulatory
framework. Conversely, many emerging market and
10
developing economies lag in AI preparedness, with
8 potential gains less than half those estimated for the
United Kingdom. This disparity stems largely from a
6
smaller proportion of workers in high-exposure and
4 high-complementarity occupations. While in advanced
economies these roles are occupied by 27 percent of
2
workers, this drops to 16 percent in emerging markets
0 and 8 percent in low-income countries. This variance
High complementarity High complementarity,
high productivity in the initial distribution of workers across occupations
reveals their reduced potential for AI benefits.
Sources: Cazzaniga and others 2024; and IMF staff For the global economy, the estimates suggest that
calculations. AI could boost productivity gains by 0.1 percent to
Note: The figure shows the change in TFP and output
between the initial and final steady state. For more details
0.8 percent annually over a decade. However, uneven
on the model, see Rockall, Pizzinelli, and Tavares 2024. distribution of these gains across regions underscores the
AI = artificial intelligence; TFP = total factor productivity. need for international cooperation to improve AI readi-
ness and integration in less-prepared nations. Initiatives
along these lines can help reduce global inequalities,
ensuring that AI benefits reach a wider array of nations.

International Monetary Fund | April 2024 83

WORLD ECONOMIC OUTLOOK—Steady but Slow: Resilience amid Divergence

References Chancel, Lucas, and Thomas Piketty. 2021. “Global Income

Acemoglu, Daron, David Autor, and Christina Patterson. 2023. Inequality, 1820–2020: The Persistence and Mutation of
“Bottlenecks: Sectoral Imbalances and the US Productivity Extreme Inequality.” Journal of the European Economic Associa-
Slowdown.” NBER Working Paper 31427, National Bureau tion 19 (6): 3025–62.
of Economic Research, Cambridge, MA. Chen, Kaiji, and Alfonso Irarrazabal. 2015. “The Role of Alloc-
Adamopoulos, Tasso, and Diego Restuccia. 2020. “Land ative Efficiency in a Decade of Recovery.” Review of Economic
Reform and Productivity: A Quantitative Analysis with Dynamics 18 (3): 523–50.
Micro Data.” American Economic Journal: Macroeconom- Coyle, Diane. 2017. “Rethinking GDP.” Finance & Development
ics 12 (3): 1–39. 54 (1): 16–19.
Adler, Gustavo, Romain A. Duval, Davide Furceri, Sinem Kılıç Crouzet, Nicholas, and Janice C. Eberly. 2021. “Intangibles,
Çelik, Ksenia Koloskova, and Marcos Poplawski Ribeiro. Markups, and the Measurement of Productivity Growth.”
2017. “Gone with the Headwinds: Global Productivity.” IMF Journal of Monetary Economics 124: S92–S109.
Staff Discussion Note 17/004, International Monetary Fund, Cugat, Gabriela, Nan Li, and Carlos van Hombeeck. 2024.
Washington, DC. “Revisiting the Relationship between GDP Growth and
Ahir, Hites, Nicholas Bloom, and Davide Furceri. 2022. “The Inequality.” Unpublished, International Monetary Fund,
World Uncertainty Index.” NBER Working Paper 29763, Washington, DC.
National Bureau of Economic Research, Cambridge, MA. David, Joel M., Hugo A. Hopenhayn, and Venky Venkateswaran.
Ahn, JaeBin, Chan Kim, Nan Li, and Andrea Manera. Forth- 2016. “Information, Misallocation, and Aggregate Productiv-
coming. “Knowledge Diffusion through FDI: Worldwide ity.” Quarterly Journal of Economics 131 (2): 943–1005.
Firm-Level Evidence.” IMF Working Paper, International Decker, Ryan A., John Haltiwanger, Ron S. Jarmin, and Javier
Monetary Fund, Washington, DC. Miranda. 2016. “Where Has All the Skewness Gone? The
Akcigit, Ufuk, and Sina T. Ates. 2021. “Ten Facts on Declining Decline in High-Growth (Young) Firms in the United States.”
Business Dynamism and Lessons from Endogenous Growth European Economic Review 86 (C): 4–23.
Theory.” American Economic Journal: Macroeconomics 13 Devries, Pete, Jaime Guajardo, Daniel Leigh, and Andrea Pesca-
(1): 257–98. tori. 2011. “A New Action-Based Dataset of Fiscal Consol-
Astinova, Diva, Romain Duval, Niels-Jakob Hansen, Ben Park, idation in OECD Countries.” IMF Working Paper 11/128,
Ippei Shibata, and Frederik Toscani. 2024. “Dissecting the International Monetary Fund, Washington, DC.
Decline in Average Hours Worked in Europe.” IMF Working Dias, Daniel A., Carlos Robalo Marques, and Christine
Paper 24/002, International Monetary Fund, Washington, DC. Richmond. 2016. “Misallocation and Productivity
Berg, Andrew, Jonathan D. Ostry, Charalambos G. Tsangarides, in the Lead Up to the Eurozone Crisis.” Journal of
and Yorbol Yakhshilikov. 2018. “Redistribution, Inequality, Macroeconomics 49: 46–70.
and Growth: New Evidence.” Journal of Economic Growth Döttling, Robin, German Gutiérrez, and Thomas Philippon.
23: 259–305. 2017. “Is There an Investment Gap in Advanced Economies?
Bils, Mark, Peter J. Klenow, and Cian Ruane. 2021. “Misallo- If So, Why?” Paper presented at the ECB Forum on Central
cation or Mismeasurement?” Journal of Monetary Economics Banking, Sintra, Portugal, June.
124: S39–S56. Duval, Romain, Gee Hee Hong, and Yannick Timmer. 2020.
Bloom, Nicholas, Charles I. Jones, John Van Reenen, and “Financial Frictions and the Great Productivity Slowdown.”
Michael Webb. 2020. “Are Ideas Getting Harder to Find?” Review of Financial Studies 33 (2): 475–503.
American Economic Review 110 (4): 1104–44. Edmond, Chris, Virgiliu Midrigan, and Daniel Yi Xu. 2015.
Budina, Nina, Christian H. Ebeke, Florence Jaumotte, Andrea “Competition, Markups, and the Gains from International
Medici, Augustus J. Panton, Marina Mendes Tavares, and Trade.” American Economic Review 105 (10): 3183–221.
Bella Yao. 2023. “Structural Reforms to Accelerate Growth, Felten, Edward W., Manav Raj, and Robert Seamans. 2021.
Ease Policy Trade-Offs, and Support the Green Transition in “Occupational, Industry, and Geographic Exposure to Arti-
Emerging Market and Developing Economies.” IMF Staff ficial Intelligence: A Novel Dataset and Its Potential Uses.”
Discussion Note 23/007, International Monetary Fund, Strategic Management Journal 42 (12): 2195–217.
Washington, DC. Felten, Edward W., Manav Raj, and Robert Seamans. 2023.
Busso, Matias, Maria Fazio, and Santiago Algazi. 2012. “(In) “How Will Language Modelers Like ChatGPT Affect
formal and (Un)productive: The Productivity Costs of Occupations and Industries?” arXiv working paper, Cornell
Excessive Informality in Mexico.” IDB Working Paper 341, University, Ithaca, NY.
Inter-American Development Bank, Washington, DC. Fernald, John G. 2015. “Productivity and Potential Output
Cazzaniga, Mauro, Florence Jaumotte, Longji Li, Giovanni before, during, and after the Great Recession.” NBER Macro-
Melina, Augustus J. Panton, Carlo Pizzinelli, Emma Rockall, economics Annual 29 (1): 1–51.
and Marina M. Tavares. 2024. “Gen-AI: Artificial Intelligence Goodhart, Charles, and Manoj Pradhan. 2020. The Great Demo-
and the Future of Work.” IMF Staff Discussion Note 24/001, graphic Reversal: Ageing Societies, Waning Inequality, and an
International Monetary Fund, Washington, DC. Inflation Revival. London: Palgrave Macmillan.

84 International Monetary Fund | April 2024

 CHAPTER 3 SLOWDOWN IN GLOBAL MEDIUM-TERM GROWTH: WHAT WILL IT TAKE TO TURN THE TIDE?

Gopinath, Gita, Sebnem Kalemli-Ozcan, Loukas Karabarbounis, Misch, Florian, and Christian Saborowski. 2018. “Resource Misallo-
and Carolina Villegas-Sanchez. 2017. “Capital Allocation cation and Productivity: Evidence from Mexico.” IMF Working
and Productivity in Southern Europe.” Quarterly Journal of Paper 18/112, International Monetary Fund, Washington, DC.
Economics 123 (4): 1915–67. Olley, Steven, and Ariel Pakes. 1996. “The Dynamics of Pro-
Gordon, Robert. 2016. The Rise and Fall of American Growth: ductivity in the Telecommunications Equipment Industry.”
The U.S. Standard of Living since the Civil War. Princeton, Econometrica 64: 1263–97.
NJ: Princeton University Press. Pattillo, Catherine A., Helene Poirson, and Luca A. Ricci. 2004.
Group of Twenty (G20). 2021. “Boosting Productivity in the “What Are the Channels Through Which External Debt
Aftermath of Covid-19.” Background Note, June 2021. Affects Growth?” IMF Working Paper 04/15, International
Ha, Doan Thi, and Kozo Kiyota. 2016. “Misallocation, Monetary Fund, Washington, DC.
Productivity, and Trade Liberalization: The Case of Viet- Pescatori, Andrea, Daniel Leigh, Jaime Guajardo, and Pete
namese Manufacturing.” Asian Development Bank Review 33 Devries. 2011. “A New Action-Based Dataset of Fiscal
(2): 94–118. Consolidation in OECD Countries.” IMF Working Paper
Hsieh, Chang-Tai, Erik Hurst, Charles I. Jones, and Peter J. 11/128, International Monetary Fund, Washington, DC.
Klenow. 2019. “The Allocation of Talent and US Economic Peters, Michael. 2020. “Heterogeneous Markups, Growth,
Growth.” Econometrica 87 (5): 1439–74. and Endogenous Misallocation.” Econometrica 88 (5):
Hsieh, Chang-Tai, and Peter Klenow. 2009. “Misallocation and 2037–73.
Manufacturing TFP in China and India.” Quarterly Journal of Pizzinelli, Carlo, Augustus J. Panton, Marina Mendes Tavares,
Economics 124 (4): 1403–48. Mauro Cazzaniga, and Longji Li. 2023. “Labor Market Expo-
Jayachandran, Seema. 2021. “Social Norms as a Barrier to Wom- sure to AI: Cross-Country Differences and Distributional
en’s Employment in Developing Countries.” IMF Economic Implications.” IMF Working Paper 23/216, International
Review 69 (3): 576–95. Monetary Fund, Washington, DC.
Jones, Charles I. 2023. “The Outlook for Long-Term Economic Restuccia, Diego, and Richard Rogerson. 2008. “Policy Distor-
Growth.” NBER Working Paper 31648, National Bureau of tions and Aggregate Productivity with Heterogeneous Plants.”
Economic Research, Cambridge, MA. Review of Economic Dynamics 11: 707–20.
Jones, Charles I., and Peter J. Klenow. 2016. “Beyond GDP? Restuccia, Diego, and Richard Rogerson. 2017. “The Causes
Welfare across Countries and Time.” American Economic and Costs of Misallocation.” Journal of Economic Perspectives
Review 106 (9): 2426–57. 31 (3): 151–74.
Khandelwal, Amit K., Peter K. Schott, and Shang-Jin Wei. 2013. Rockall, Emma, Carlo Pizzinelli, and Marina Mendes Tavares.
“Trade Liberalization and Embedded Institutional Reform: 2024. “Artificial Intelligence Adoption and Inequality.” Unpub-
Evidence from Chinese Exporters.” American Economic Review lished, International Monetary Fund, Washington, DC.
103 (6): 2169–95. Song, Zheng, Kjetil Storesletten, and Fabrizio Zilibotti. 2011. “Grow-
Kilic Celik, Sinem, M. Ayhan Kose, and Franziska Ohnsorge. ing like China.” American Economic Review 101 (1): 196–233.
2023. “Potential Growth Prospects: Risks, Rewards, and Syverson, Chad. 2017. “Challenges to Mismeasurement
Policies.” Policy Research Working Paper 10355, World Bank, Explanations for the US Productivity Slowdown.” Journal of
Washington, DC. Economic Perspectives 31 (2): 165–86.
Levy, Santiago. 2018. Good Intentions, Bad Outcomes: Social Pol- Tombe, Trevor, and Xiaodong Zhu. 2019. “Trade, Migration,
icy, Informality, and Economic Growth in Mexico. Washington, and Productivity: A Quantitative Analysis of China.” Ameri-
DC: Brookings Institution Press. can Economic Review 109 (5): 1843–72.
Libert, Thibault. 2017. “Misallocation before, during and after the Ulyssea, Gabriel. 2018. “Firms, Informality, and Development:
Great Recession.” Working Paper 658, Bank of France, Paris. Theory and Evidence from Brazil.” American Economic Review
Midrigan, Virgiliu, and Daniel Yi Xu. 2014. “Finance and 108 (8): 2015–47.
Misallocation: Evidence from Plant-Level Data.” American Woo, Jaejoon, and Manmohan S. Kumar. 2015. “Public Debt
Economic Review 104 (2): 422–58. and Growth.” Economica 82 (328): 705–39.
Milanovic, Branko. 2002. “True World Income Distribution, World Bank. 2022. “Impact of COVID-19 on Global Income
1988 and 1993: First Calculation Based on Household Sur- Inequality.” In Global Economic Prospects: January 2022,
veys Alone.” Economic Journal 112: 51–92. 155–200. Washington, DC: World Bank.

International Monetary Fund | April 2024 85