Understanding Economic Growth: Key Factors and Insights
School
London School of Economics**We aren't endorsed by this school
Course
EC 2B1
Subject
Economics
Date
Dec 11, 2024
Pages
62
Uploaded by MateDangerArmadillo40
1EC2B1: MACROECONOMICSWeek #2A Simple Framework to Think about GrowthWouter Den Haan
2Required Reading•Gordon (2012): Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six Headwinds.•Section 3.5 in Jones and Vollrath is also on growth accounting, but better to only read this after we have covered the Solow growth model in continuous time.
3Assignment #2•Use real world data to practice the growth-accounting tool learned in the lecture.•The lecture showed that a necessary condition for convergence is that poor countries grow faster. The assignment deals with the question whether this is also a sufficient condition.•Python Skills: By programming a growth-accounting exercise you will not only learn a bit of basic programming, but also gain a solid understanding of growth accounting.
4Overview1.What affects GDP?2.Framework to think about factors determining GDP.a.Employment.b.(Different types of) capital.c.Productivity (broadly defined).3.Use framework to understand:a.Which factors are behind differences across countries?b.Which factors are behind differences across time?4.Two case studies:a.Hong Kong and Singapore.b.Bangladesh.5.Current headwinds: US and UK economic growth.
5What is best way to increase GDP? 1.Stimulate consumption (firms will hire more workers).2.Stimulate investment (production capacity will increase).
6What affects GDP?
7What affects GDP?1.Labour.2.Capital.3.Productivity/efficiency ≡Total Factor Productivity (TFP).
8How Does Labour Affect GDP?•If more hoursare put into producing output ⇒GDP ↑. •What affects the labour input?•Population.•Labour force participation.•Employment rate.•Hours worked per employed person.
9How Does Capital Affect GDP?•Capital is a tool ⇒Capital ↑ means that GDP ↑. •There are different types of capital:•Physical.•Intangible (also referred to as intellectual capital/property).•Human capital,•Relationship capital,•Structural capital: processes, data systems that are helpful in running companies efficiently.
10How Does Productivity Affect GDP?If the two inputs, capital and labour,are used more efficiently⇒productivity ↑ & GDP ↑. What affects productivity at the aggregate level?•Technology.•Regulation.•Allocation across different uses (firms).•Efficiency of usage (management, work ethic).•Capacity utilization.
11Productivity vs Intangible Capital•Productivity and intangible capital seem similar concepts.•They are!•Strictly speaking, acquiring capital requires investment, whereas increased productivity may be possible through•Learning by doing,•Copying ideas from others.•But in some productivity theories increases occur because of investmentin research and development (R&D).•For the theories we consider in this course: Think of capital as physical capital.
12Poor countries are poor because …a.Poverty trap.b. Geography.c.Low investment in physical capital.d.Low investment in human capital.e.Bad institutions.f.Something else.
LSE Faculty ProfileProfessor Caselli has written numerous articles on economic growth. That includes some big-picture papers explaining the main drivers behind economic growth like we do in this lecture.But also on specific questions such as computer-technology adoption. One very intriguing outcome of his research on resource windfalls is that those can be bad for the country!!! For example, they can make an autocratic regime even more autocratic. Francesco Caselli (webpage)Norman Sosnow Professor of Economics
Source: Caselli (2016)Massive Income Differences - 2005
Source: Caselli (2016)Includes natural capital (i.e. natural resources)Large Differences in Capital - 2005
Source: Caselli (2016)Note: Human capital = schooling + health… and human capital 2005
Source: Caselli (2016)Note: Years of total schooling (including higher educ.)Schooling Matters a Lot - 2005
18Let’s create some structure in our thinking
19Simple Framework: GDP࠵? = ࠵?!(࠵?࠵?)"#!,0 < ࠵? < 1•࠵?:GDP•࠵?: Capital•࠵?: Employment (persons working)•࠵?: Total Factor Productivity
20Simple Framework: GDP࠵? = ࠵?!(࠵?࠵?)"#!,0 < ࠵? < 1Properties:•Constant Returns to Scale (CRS) with respect to all inputs. That is, (࠵?࠵?)!(࠵?࠵?࠵?)"#!= ࠵?!࠵?"#!࠵?!(࠵?࠵?)"#!= ࠵?࠵?, where ࠵?is an arbitrary number.•Decreasing Returns to Scale (DRS) with respect to one input:•For fixed Aand L: ∆%∆&and ’%’&decrease with ࠵?.•DRS is also referred to as diminishing marginal product
21TFP & GDP: 3 specifications•Neutral: ࠵? = ࠵?࠵?(࠵?, ࠵?)•Labour-augmenting (or labour deepening): ࠵? = ࠵?(࠵?, ࠵?࠵?)•Capital-augmenting (or capital deepening): ࠵? = ࠵?(࠵?࠵?, ࠵?)
22TFP & GDP: 3 specifications•When ࠵?(-)is a Cobb-Douglas production function, then there is no difference between the three.•For example, a production function with labour-augmenting technological progress can be rewritten as one with neutral progress.࠵? = ࠵?!࠵?࠵?"#!= ࠵?"#!࠵?!࠵?"#!=.࠵?࠵?!࠵?"#!,where .࠵? = ࠵?"#!.•Thus, the labour-augmenting production function has turned into one with neutral technological progress.
23Simple Framework: GDP per worker࠵?: GDP per (working) person࠵?: Capital per (working) person࠵? =࠵?࠵?=࠵?࠵?!࠵?"#!࠵?!࠵?"#!= ࠵?࠵?࠵?!࠵?࠵?"#!= ࠵?࠵?!Thus,࠵? = ࠵?࠵?!•DRS for GDP per worker!•Data is typically “GDP per capita” not “GDP per working person!”•This does not affect insights below as long as ࠵?as a fraction of population remains constant. And you shouldn’t worry about that in this course.
yakaGDPPer Capita Capital StockkbybkcycüSame ࠵?࠵?üDifferent ࠵?࠵?Diminishing Returns(also referred to as diminishing marginal product)==
25Small Extensionℎ: measure to indicate either human capital or hours worked࠵? = ࠵?!(࠵?ℎ࠵?)"#!࠵? =࠵?࠵?=࠵?!(࠵?ℎ࠵?)"#!࠵?!࠵?"#!=࠵?࠵?!࠵?ℎ࠵?࠵?"#!= ࠵?!(࠵?ℎ)"#!•Factors of production:࠵?!ℎ"#!.•Output per worker = productivity ×factors of production.•Do we still have DRS for GDP per worker?Yes, since h(and A) do not automatically go up with k
26What questions can we answer with our framework?
27Key QuestionsDevelopment/growth Accounting1.Is it A, K, or hL when comparing differences across countries?2.Is it A,K, or hLwhen we compare GDP growth for a specific country over time?
28Measuring Productivity•!!! ࠵?is not observed,but ࠵?, ࠵?, and ℎare observed. And we have some idea about ࠵?.•Then you can obtain ࠵?from ࠵?"#!=$%!&"#!.
29Calculating Level Differences•Upcoming slides show results from Caselli (2016).•Getting the data and making sure they are comparable across countries is difficult.•Caselli assumes neutral technological progress, but we know that doesn’t matter for Cobb-Douglas.࠵? = ࠵?࠵?!ℎ"#!= ࠵?࠵?!ℎ"#!= ࠵? 5࠵?
30Calculating Level Differences•Suppose we have ࠵? = 0.138࠵?’(and ℎ = 0.66ℎ’(for a particular income decile.•Then we get for the combined role of both types of capital5࠵?5࠵?’(=࠵?!ℎ"#!࠵?’(!ℎ’("#!=(0.138࠵?’()!(0.66ℎ’()"#!࠵?’(!ℎ’("#!= 0.138!0.66"#!= 0.3938•That is, ࠵? < ࠵?’(and ℎ < ℎ’(means a 39.38% lower value for 5࠵?(and a 39.38% lower value for yif ࠵? = ࠵?’().
31Calculating Level DifferencesAnd the role for TFP is the residual. That is,࠵?࠵?’(=࠵?/5࠵?࠵?’(/5࠵?’(=࠵?/࠵?’(0.3938For example,•if ࠵? = 0.3938࠵?+,, then ࠵? = ࠵?+,.that is, the difference between yand ࠵?+,is due only to the two types of capital. •If ࠵? < 0.3938࠵?+,, then ࠵? < ࠵?+,.•For example, if ࠵? = 0.165࠵?+,then ࠵?/࠵?+,=0.165/0.3938 = 0.4190. That is, ybeing only 16.5% of࠵?+,is because Ais 41.9% lower and ࠵?!ℎ"#!is 39.38% lower.
Source: Caselli (2016)Documenting Level Differences: 2005This is what needs to be explained(the average country in the 10thdecile is slightly richer than the US)1.63.76.510.916.522.831.457.478.8103.902040608010012345678910deciledecile averageIncome per Worker relative to USA (percent)Figure 1: World Income Distribution
Source: Caselli (2016)Documenting Level Differences: 2005The richest countries have more physical capital per worker than the US 2.43.47.010.813.820.929.066.398.7152.205010015012345678910income decileaverage by income decilePhysical Capital per Worker relative to USA (percent)Figure 2: Endowments of Physical Capital
Source: Caselli (2016)Documenting Level Differences: 200549.354.859.670.166.671.875.683.983.782.802040608012345678910income decileaverage by income decileHuman Capital per Worker relative to USA (percent)The richest countries have less human capital per worker than the US
Source: Caselli (2016)Documenting Level Differences: 2005Conclusion: both capital (physical and human) and TFP matter02040608010012345678910income decileaverage by income decileTotal capital and Efficiency Relative to USA (percent)total relative capitalrelative efficiencyFigure 9: Baseline Results
36GROWTH ACCOUNTINGStart with࠵?!= ࠵?!࠵?!"࠵?!#$"Take logs ⇒ln ࠵?#= ln ࠵?#+ ࠵? ln ࠵?#+1 − ࠵? ln ࠵?#ln ࠵?%= ln ࠵?%+ ࠵? ln ࠵?%+1 − ࠵? ln ࠵?%⇒ln ࠵?%− ln ࠵?#= ln ࠵?%− ln ࠵?#+ ࠵?(ln ࠵?%− ln ࠵?#) +1 − ࠵? (ln ࠵?%− ln ࠵?#)⇒ln࠵?)࠵?"= ln࠵?)࠵?"+ ࠵? ln࠵?)࠵?"+ (1 − α)ln࠵?)࠵?"
37GROWTH ACCOUNTING•ln*$*"is the growth rate ofX. (See homework)•Thus, we get࠵?+= ࠵?,+ ࠵?࠵?-+ (1 − ࠵?)࠵?.1 =࠵?,࠵?++ ࠵?࠵?-࠵?++ (1 − ࠵?)࠵?.࠵?+
38GROWTH ACCOUNTING•Fraction of output growth due to TFP:/%/&•Fraction of output growth due to capital accumulation:!/’/&•Fraction of output growth due to employment growth:("#!)/(/&
39GROWTH ACCOUNTING•Note that the three fractions must add up to 1.•In practice, you use data onY,K,andLto get a time series forA.That is,Ais a residual not a direct measure of technology.•It doesn’t matter whether you use the neutral or the labour-augmenting version. You get the same three numbers. (seehomework).
40Documenting Growth Differences: 1975-2009Growth rate of Factors of ProductionGrowth rate of TFPPoorest quintile0.44%-1.42%Second quintile0.89%-0.55%Third quintile0.95%0%Fourth quintile1.16%0.35%Richest quintile1.87%1.33%Source: Weil (2013, Chapter 7)Conclusion: Again both TFP and factors of production matter
41A Tale of Two CitiesHong Kong & SingaporeBased on Young (1992)
LSE Faculty ProfileAlwyn Young is an expert on economic growth and international trade. He has developed models of endogenous growth and several in-depth papers on the growth performance of specific countries and regions. His research on the consequences of the AIDS tragedy shows that orphaned children are disadvantaged in terms of experiencing lower human capital accumulation, but benefit later on because of the induced scarcity of labour. In terms of consumption levels, the latter effect dominates.Alwyn Young (webpage)Professor of Economics
43•Both used to be British colonies.•Trading ports with initially little manufacturing.•Both developed export-dependent manufacturing sector (textiles → clothing → plastics → electronics).•Eventually, also banking and financial services.•High growth and high investment rates. Hong Kong & Singapore: Similarities
44•Initially, Hong Kong residents better educated.•Hong Kong: Laissez-faire economic policy.•Singapore: More authoritarian and forced national savings.Hong Kong & Singapore: Differences
45Hong Kong & Singapore: Growth AccountingGrowth of Average Percentage contribution of Time capital period Output Labor Capital share Labor Capital TFP A Hong Kong 61-66 0.577 0.130 0.694 0.393 0.14 0.47 0.39 66-71 0.322 0.126 0.377 0.355 0.25 0.42 0.33 71-76 0.406 0.098 0.361 0.330 0.16 0.29 0.54 76-81 0.512 0.350 0.527 0.386 0.42 0.40 0.18 81-86 0.294 0.108 0.374 0.421 0.21 0.54 0.25 Singapore 66-70 0.507 0.157 0.576 0.562 0.14 0.64 0.23 70-75 0.454 0.317 0.860 0.553 0.31 1.05 -0.36 75-80 0.408 0.289 0.466 0.548 0.32 0.63 0.05 80-85 0.300 0.249 0.474 0.491 0.42 0.78 -0.20 Source: Young (1992, Table 6)
46Alwyn Young’s Explanation
47Taking Stock•∆࠵?can be important for growth even if ∆࠵?࠵?࠵?is not.•Obvious questions:•Can we model how ∆࠵?affects growth?•Is this growth through ∆࠵?sustainable?•Solow modelprovides answers.
48Bangladesh: A recent success story?
49Bangladesh: Key obs. for the Tiger Cub•Population of 165 million.•1971 Independence.•1971-1989: 1.3% growth of GDP per capita.•1995-2015: 5% growth of GDP per capita.•1991: 43.7% < poverty line.•2016: 14.5% < poverty line.•Politically unstable.•Although Sheikh Hasina (daughter of murdered previous head of state) has consolidated power since 2009.•Clearly NOT without corruption•Ranked 147 on the Transparency International “Corruption Perceptions Index.”•Clientelist regime.
50Bangladesh: Key Factors•FDI in Garment Industry (bit of luck, but it did have favourable export environment).•Lots of remittances (5 million Bangladeshis live abroad).•Structural transformation (manufacturing share in GDP went from 12% in 1991 to 21% in 2017).•Improved health care and education.•Financial development in the form of micro credit.•Grameen bank (Nobel Peace Prize winner 2006).•Growth may have required exploitation of workers! But “The misery of being exploited by capitalists is nothing compared to the misery of not being exploited at all” according to the Marxist economist Joan Robinson.
51Success in presence of corruptionSuccess in business•Dercon (2022, p. 223): “Throughout this period, the state was weak, dysfunctional, and often unstable and violent.”•BUTstill focused on growth being not predatory and excessively rent seeking. The “elite” does at times understand that long-termgrowth is beneficial for them as well.Success in health and education:•Bangladesh did not curtail grass root organizations (often financed with foreign financial support) as long as these kept out of high-level politics.Dercon (222, p. 219): “Progress is not at all the result of a grand design. Instead, it is, at most coincidentally, a matter of politics and economics not doing the wrong thing.”
52Current Headwinds to UK and US Economic Growth
53US GDP TREND GrowthKey observation: Long-term trend growth dropped before financial crisis. Posterior Median estimate and 68% and 90% credible intervals.From: Antolin-Diaz, Drechsel, Petrella (2017)
54Headwinds for the US (i.e., the leader)based on Gordon (2012)•Previous high growth was due to many follow up inventions of the 2ndIndustrial Revolution. •The possibilities of the 3rdIndustrial Revolution are not quite as spectacular, but innovations are hard to predict. Also,1.No more further growth because of increased female labour force participation.2.Fraction of higher education people unlikely to increase a lot more.3.Rising inequality•Higher averageGDP per capita may mean higher income for just a few.•Inequality may itself create problems, e.g., political polarization.4.Competition with low-wage countries.5.Energy and the environment.6.High government debt levels. •(Household debt levels no longer very high but corporate debt levels are).
55UK Relative PerformanceSource: Economist, June 11 2022
56UK Relative PerformanceSource: Economist, June 11 2022Also, see the 2024 report of the Institute for Public Policy Research [link]
57Additional UK observations1.The finance and manufacturing sector can account for most of the fall in UK aggregate productivity. See Tenreyro (2018).2.The problem seems to be that most productive firms see less growth than what they used to accomplish. See Schneider (2018).Silvana Tenreyro is a professor at the LSE. From 2017 to 2023, she also was a member of the Bank of England’s Monetary Policy Committee. Her research has focused on many different fields in macroeconomics, including growth, monetary policy, and housing to name just a few.
58What is the Main Reason for low UK (Productivity) Growth?1.Reduced profitability financial sector especially important for UK.2.Hard to get permits to expand businesses.3.Housing very expensive.4.Financial sector not great in financing young innovative firms.5.Regional inequality and inefficient allocation of resources.6.Education levels not that great.
59Reasons for poor UK performance1.Reduced profitability financial sector especially important for UK.2.Hard to get permits to expand businesses.3.Housing very expensive.4.Financial sector not great in financing young innovative firms.5.Regional inequality and inefficient allocation of resources.6.Education levels not that great (32% of population has higher secondary qualification relative to 42% OECD average).
60So which one is the culprit?Answer: We do not really know!Diane Coyle may be right when she describes this as an Agatha Christie mystery: “Everybody turns out to have done it.”
61Key Lessons1.There is a wide variety of growth patterns, both across time and space.2.But both differences in capital (physical and human) and efficiency (TFP) matter.3.Sustained growth is a bit of a miracle and there are some headwinds.
62References &Additional Resources•Tracking the Slowdown in Long-Run GDP Growth•With Juan Antolin-Diaz and Ivan Petrella -Review of Economics and Statistics, Vol. 99(2), May 2017.•Antolin-Diaz, J., T. Drechsel, and I. Petrella, 2017, Tracking the Slowdown in Long-Run GDP Growth, Review of Economics and Statistics, 99(2).•Caselli, F., and J. Feyer, 2007, The Quarterly Journal of Economics, May pp. 535-568.•Caselli, F., 2016, Accounting for Cross-Country Income Differences: Ten Years Later, World Development Report.•Dercon, S., 2022, Gambling on Development: Why Some Countries Win and Others Lose.•Schneider (2018): The UK’s productivity puzzle is in the top tail of the distribution.•Tenreyro (2018): The fall in productivity growth: Causes and implications.•Young, A., 1992, A Tale of Two Cities, NBER Macroeconomics Annual 7, pp. 13-54.