1 “Macroeconomics” of Development

2 Plan of Action Course goals & logistics Historiography of developmentPer-capita income Understanding long-run growth Convergence/divergence in the “modern” era Understanding “modern” PCI levels/growth Proximate causes (growth accounting) and the search for “deeper” causes

3 Course Goals Articulate the research frontierDevelop the capacity to be a critical consumer of applied development research (Not a tools class) Prerequisites: 1st year PhD theory and metrics sequence Econ 250A (labor methods)

4 Topical Coverage Proximate causes of growth: factors of productionPhysical capital Human capital / technology Generic impediments Complementarities & poverty traps Institutions Corruption

5 A (Brief) Historiography of Development

6 A Brief Historiography of DevelopmentGrowth has always been a priority question… WoN: source of wealth is productivity, not gold …but focus was on rich countries Post-WWII independence wave created new intellectual space Urgent demand for cross-applicable ideas Space for idea entrepreneurs

7 Three Ingredients Keynesian intellectual climateThe west had recently emerged from the Great Depression Perceived success of the USSR via forced investment Pessimism about markets in LDCs Lewis (1954) “Economic Development with Unlimited Supplies of Labor”

8 Lewis (1954) “The classics, from Smith to Marx, all assumed, or argued, that an unlimited supply of labour was available at subsistence wages. They then enquired how production grows through time. They found the answer in capital accumulation…” “In the first place, an unlimited supply of labour may be said to exist in those countries where population is so large relatively to capital and natural resources, that there are large sectors of the economy where the marginal productivity of labour is negligible, zero, or even negative… …twenty years ago one could not write these sentences without having to stop and explain why in these circumstances, the casual labourers do not bid their earnings down to zero, or why the farmers‘ product is not similarly all eaten up in rent, but these propositions present no terrors to contemporary economists.”

9 Dominant frameworks Harrod-Domar frameworkCreated to study business cycles, reinterpreted as a growth theory: growth proportional to net investment Rosenstein-Rodan (1943) “Problems of Industrialization of Eastern and South-Eastern Europe” (the “big push”) Rostow’s “Stages of Growth” (the “takeoff into sustained growth”)

10 Consequences For policy For the field as an intellectual endeavorHeavy emphasis on state-led capital investment and infant- industry protection Disappointment and backlash – the Washington Consensus For the field as an intellectual endeavor [-] Tremendous amounts of work on “priority sectors” for investment that has largely been forgotten [+] Placed development at the forefront of debates on rationality and market efficiency E.g. moral hazard (sharecropping, efficiency wages), adverse selection (credit rationing) Schultz (1964) “Transforming Traditional Agriculture” – the rational peasant

11 Development Economics TodayPessimistic view Abdicated the big picture Optimistic view Broader appreciation of proximate causes of growth beyond capital (human capital, technology) and active work on the micro-economics of these factors: from “which sectors?” to “what frictions?” More realistic, endogenous view of the state Better empirical tools

12 Understanding Long-run Growth

13 Per Capital Income “Modern” growth is a recent phenomenonDeLong, “Slouching Towards Utopia”

14 Lifestyles, Then and NowIf anything our income statistics likely understate modern changes in living standards due to reference basket problem Thought experiment: how much 1900 income would you accept to give up treatments for heart disease/cancer/polio, cell phones, laptops, internet, movies, television, etc.? In contrast, great continuity prior to 1800s Clark (1957) on Ancient Rome v.s. 18th century England: e.g. home heating technologies

15 Understanding Long-Run GrowthFocus has been on simultaneously explaining two questions: Why was growth so slow for so long? Why did growth accelerate when it did? Answers have emphasized interplay with fertility Malthusian dynamics Innovation (Kremer 1993) Women’s time (Galor & Weil 1996)

16 Malthus vs. Kremer Land is fixed so only ideas can raise incomeMalthus: fertility adjusts to fix Y/p at some subsistence income Kremer: technological growth is proportional to number of heads

17 Implications Population growth proportional to population levelsExplosive growth of Y But not y

18 Explaining the TransitionIf fertility is not infinitely income-elastic, will get some PCI growth from this model (fertility responds to income more slowly than income to fertility) Empirically, fertility levels are non-monotonic in PCI (key non- Malthusian ingredient) Combining these yields threshold behavior and a growth takeoff

19 Refinements Technology and scale Fertility and the transitionDiamond “Guns, Germs and Steel” Geography of endowments: the incorrigible zebra problem Geography of diffusion: East/West v.s. North/South major continental axes Fertility and the transition E.g. Galor & Weil (1996): productivity growth affects the relative price of women’s time

20 Alternative PerspectivesNorth (1989, 1990): institutions Increasing costs of medieval warfare Monarchs traded commitments to protect property and enforce contracts for revenues Specific technology view Abstract – Scientific Method Concrete – steam engine

21 The “Modern” Era General consensus that recent (150-year) history has been qualitatively different Additional watershed in the 1950s: independence How relevant is work on the transition / Industrial Revolution today? One view: LDCs today are like England in 1700s and must follow the same path Alternative view: being poor in a rich world is very different (e.g. cellphones) Much commonality – e.g. pro-innovation institutions may also promote adoption

22 Convergence / Divergence in the Modern Era

23 PCI Across Countries Barro & Sala-i-Martin 2004

24 Sala-i-Martin (2006)

25 Questions about the WDIAre we converging or diverging? Theory: mechanisms Evidence Questions to ask What does this imply for the future of LDCs? What markets do we need to understand? What policy levers affect convergence/divergence?

26 Mechanisms Autarky GlobalizationDiminishing returns (Solow) v.s. increasing returns (Romer) Globalization Geopolitical interaction (today) Capital flows (later) Knowledge flows (later)

27 Solow (1956) Contrasted with prevailing contemporary Harrod-Domar view – Leontief production Growth rate of capital per worker is decreasing in its level, which probably implies convergence

28 Solow (1956) When are growth rates fastest?Empirically the capital share seems fairly stable over time (a “Kaldor fact”) Cobb-Douglas f() does exactly this Implies convergence In the sense of a [-] relationship between initial income levels and subsequent growth rates Continues to hold with exogenous tech. progress Need not imply falling s.d. of log(y)

29 Aside: Kaldor (1963) facts Per capita output grows at a roughly constant rate Physical capital per worker grows over time The rate of return to capital is roughly constant The ratio of physical capital to output is nearly constant The shares of labor and physical capital in national income are nearly constant Growth of output per worker varies substantially across countries.

30 Romer (1986) Increasing returns is a venerable ideaSmith: division of labor (pin factory) Arrow: learning by doing Little work had been done on the topic for technical reasons Internal increasing returns tend to yield multiple, awkward equilibria, e.g. single-firm economies Idea: emphasize Marshall’s external economies

31 Romer (1986) External increasing returnsAre compatible with competitive equilibrium Can generate explosive / divergent growth

32 How does globalization affect inequality?Geopolitical interactions Institutional diffusion Colonialism (Neocolonialism: “core” and “periphery”) Capital flows Could work either way depending on how MPK varies with K (Solow v.s. Romer, Lucas) Knowledge flows Presumably more one-sided, but also harder to document May depend on “specificity” or “appropriateness”

33 Geopolitical InteractionsInstitutional diffusion Technology diffusion on a national scale US “laboratory of democracy” -> DiDs Mukand & Rodrik 2005: national similarity and adoption Colonization Some former colonies are now rich (US, Canada, Australia) Some former colonies are now poor (Africa) Was colonialism a force for convergence or divergence? Between colonizers and colonized:? Diamond: initial income gap -> colonization -> bigger income gap. Among colonized?

34 Acemoglu, Johnson, Robinson (2002)

35

36 Interpretation Geography? ColonizationFactors that favored early development may have hindered subsequent growth Colonization Colonial powers faced a choice of institutions Extractive institutions: heavy tax rates, forced labor Pro-market institutions: contract enforcement, property rights, conducive to settlement We will return to these ideas later in the course Argue that timing & sensitivity to controls more consistent with the latter

37 Evidence on ConvergenceConvergence is an empirical question Unconditional Convergence Baumol (1986) v.s. Pritchett (1997) Conditional Convergence Barro & Sala-i-Martin (1992) Household-level data Sala-i-Martin (2006)

38 Baumol (1986) Data on PCI are a key constraintData on 16 countries for from Maddison (1982) Data on 72 countries for from Summers & Heston (Penn World Tables)

39 GDP Per Work Hour

40 Growth v.s. Initial Level

41 Interpretation “It seems not to have mattered much whether or not a particular country had free markets, a high propensity to invest, or used policy to stimulate growth. Whatever its behavior, that nation was apparently fated to land close to its predestined position in Figure 2.” “In other words… a successful productivity-enhancing measure has the nature of a public good. And because the fruits of each industrialized country's productivity-enhancement efforts are ultimately shared by others, each country remains in what appears to be its predestined relative place along the growth curve of Figure 2. I will note later some considerations which might lead one to doubt that the less developed countries will benefit comparably from this sharing process.”

42 Summers & Heston data

43 Interpretation “This suggests that there is more than one convergence club. Rather, there are perhaps three, with the centrally planned and the intermediate groups somewhat inferior in performance to that of the free-market industrialized countries. It is also clear that the poorer less developed countries are still largely barred from the homogenization processes… part of the explanation may well be related to product mix and education.”

44 Pritchett (1997) Critique Alternative approachCountries in Maddison’s data are rich now In the past they were either rich or poor This in itself implies convergence Alternative approach Missing historical data for the now-poor is the key constraint Bound GDP per capita below by subsistence levels, estimated at $250 / year

45

46 Pritchett – interpretation“Taken together, these findings imply that almost nothing that is true about the growth rates of advanced countries is true of the developing countries, either individually or on average. The growth rates for developed economies show convergence, but the growth rates between developed and developing economies show considerable divergence. The growth rates of developed countries are bunched in a narrow group, while the growth rates of less developed countries are all over with some in explosive growth and others in implosive decline.”

47 Can We Do Better than “Clubs”?Can always identify convergent subsets ex-post Do our theories deliver ex-ante predictions about who should converge? Solow: similar rates of saving, population growth; shared technological progress More generally one might think of institutions (property rights, provision of public infrastructure) Need a set of units that are similar in these dimensions

48 Barro & Sala-i-Martin (1992)Cobb-Douglas version of NGM implies Key assumptions Units have been somehow perturbed to differ in initial income Units share a common tech. growth process Countries probably don’t satisfy these, but US states may How does the Civil War factor in? What are potential sources of bias & which way do they push? Unobserved heterogeneity? Measurement error in initial income? US states autarkic?

49

50

51 , 98 countries

52 Remarks Cross-country results flip with enough controlsConsistent with “conditional convergence” But controls are ad hoc and their exogeneity questionable (e.g. contemporaneous G/GDP?) See Mankiw, Romer, Weil (1992) for a more structured approach to conditional convergence OECD results are intermediate

53 Sala-i-Martin (2006) Literature focuses on distribution of PCI across countries Correct for testing growth theories driven by factors that are mobile within countries General result has been divergence Potentially misleading for understanding human well-being. Do we care less about a Chinese peasant than a Senegalese farmer because he lives in a bigger country?

54

55

56 Sala-i-Martin (2006) Population weights are a step in the right direction, but could also be very misleading for measuring poverty (e.g.) – why? How to measure world distribution of per-capita income? Nationally representative household surveys are only available for some countries and years Approach: Use annual national accounts to estimate average PCI in each country/year Use most recent survey data to impute distribution around that mean

57

58 Sala-i-Martin (2006) Rising inequality within countriesRising inequality across countries And yet falling inequality across households India & China

59 Explaining PCI and GrowthVarious forms of “explanation” Growth/levels accounting What observables let me accurately predict income/growth? Can sometimes be though of as “proximate causes” Causal What can I change to change growth/income? Runs up against deep philosophical issues about locus of agency

60 Understanding Modern Growth

61 Proximate v.s. “deep” causesWhy did Lincoln die? Proximate: JWB shot him Deeper: JWB’s childhood, attendance at John Brown’s execution, etc. Why do countries grow?

62 Statistical & Accounting ExplanationsDoes the NGM provide a good fit? Countries may share a common technology frontier In this case, PCI differences simply reflect differences in accumulable factors Which in turn reflect differences in savings, fertility, etc. Pushes us towards understanding concrete forms of investment (physical and human capital) rather than technology Hotly debated Yes: Mankiw, Romer, Weil 1992; Islam 1995; Young 1995 No: Klenow & Rodriguez-Clare 1997; Hsieh 2002

63 Mankiw, Romer, Weil (1992) Common assertion: Solow model explains cross-country income differences by appealing to black-box A Though Solow did not call it a theory of international differences. Is this true? Other parameters: savings rate, population growth rate, depreciation rate With what would you replace savings rate in the Ramsey-Cass- Koopmans NGM?

64 Base Solow Model Constant returns overallDecreasing returns to accumulable factors Exogenous rates of saving, population growth, technological progress, and depreciation (Small-caps are per effective unit of labor)

65 Predictions Suppose many countries with access to the same technology and experience the same depreciation rate In steady-state, all grow at common rate g Do we care about the steady state? Does this make the test more or less convincing? This leaves variation in savings and fertility to explain cross-country differences Estimation equation:

66 Choice of Tests Solow model also impliesWhy not estimate this instead? Pros? Cons? Could also fix alpha using factor shares data rather than estimating it – pros / cons of this approach? Pro: does not assume we are in steady-state, don’t need to guess a depreciation rate Con: capital stock is much harder to measure than savings / fertility rates

67 Assume s and n are exogenousHow much of cross-country income variation can savings and population growth explain (statistically)? Data: 98 Summers-Heston countries and subsets (now Penn World Tables)

68 “Non-structural” takeCountries that invest more are richer Countries whose populations grow faster are poorer Surprising part is the high R2 Hinges on an estimated value for the capital share of 0.59 (for intermediate sample) much higher than the 0.3 usually assumed Huge parts of investment are in human capital; is this an important missing piece?

69 Augmented Solow Model Low-caps are per effective unit of laborAssumptions α + β < 1 Physical capital and human capital produced using same p.f. and depreciate at the same rate

70 Steady-State Definition: growth in capital and human capital per effective unit of labor halts (Why do steady-state levels of one factor depend on investment in the other?) Growth in output per worker continues at the exogenous steady-state rate g

71 A Broad View of Capital Adding human capital implies higher sensitivities Measuring sh Ideal: % of income spent on human capital Proxy: fraction of working-age population in school What are some issues with this measure? Stocks v.s. flows Quality v.s. quantity of schooling Non-schooling investments in self Which way does this bias the test?

72 How does measurement error effect interpretation?Higher R2 How does measurement error effect interpretation? Error in LHS? Error in RHS?

73 Islam (1995) MRW specification assumes investment rates and population are exogenous Investment clearly response to “technology” (ideas, infrastructure, institutions, etc.) If these are time-invariant then a country FE estimator removes the bias

74 This is a panel data model with a lagged dep. var.Estimation? Note the symmetry restriction implied by theory Caveats Removing more of the “good” variation or of the “bad”? Lagged dependant variable problems: LSDV requires large T Islam provides finite-sample Monte Carlos and an alternative MD estimator

75 Strikingly good fits More realistic estimates of the capital share

76 Young (1995) “Asian Tigers” seen as models of successful development“This is a fairly boring and tedious paper, and is intentionally so. This paper provides no new interpretations of the East Asian experience to interest the historian, derives no new theoretical implications of the forces behind the East Asian growth process to motivate the theorist, and draws no new policy implications from the subtleties of East Asian government intervention to excite the policy activist. Instead, this paper concentrates its energies on providing a careful analysis of the historical patterns of output growth, factor accumulation, and productivity growth in the newly industrializing countries (NICs) of East Asia, i.e., Hong Kong, Singapore, South Korea, and Taiwan.” “Asian Tigers” seen as models of successful development Debates Market orientation or state leadership? Productivity growth or factor accumulation?

77 Painstaking collection of data on accumulation of physical and human capitalResult: not much left over

78 Klenow & Rodriguez-Clare (1997)“In our view these studies constitute a neoclassical revival. They suggest that the level and growth rate of productivity is roughly the same across countries, so that differences in output levels and growth rates are largely due to differences in physical and human capital. Romer (1993), in contrast, argues that "idea gaps" are much more important than "object gaps." In terms of a simplified production function Y = AX, where A is productivity and X encompasses physical and human capital, this debate is over the relative importance of A and X.” “This debate matters because the positive and normative implications of the A view can differ dramatically from those of the X view. Technology-based models of A exhibit scale effects because of the nonrival nature of technology creation and adoption. And they suggest that openness, perhaps though its effect on technology diffusion, can have first-order effects on living standards and growth rates (without requiring big differences in rates of return to capital)… These implications are not shared by the basic neoclassical growth model, which has the same technology everywhere.”

79 Handling Human CapitalHigh R2’s in MRW hinge on two things High estimated sensitivity of y to h Extensive variation in h Sensitivity OLS will choose sensitivity that minimizes MSE – like maximizing R2 We can potentially do better by fixing these at plausible values, guided by micro-evidence on returns to human capital Also partially sidesteps the endogeneity concerns raised by Islam Variation MRW use highly-variable secondary schooling rates Primary schooling rates vary much less

80 Fixed coefficients at MRW estimates: (α = 0.30,β = 0.28) Columns: contributions of capital, human capital, and both to explaining 1985 PCI Rows MRW0: replication exercise MRW1: removes HC investment from Y MRW2: updates dataset MRW3: uses Barro-Lee (1993) enrollment data to form HY as all enrollment (primary + secondary) MRW4: uses preferred estimates of education production function parameters

81 Growth version As in Islam (1995), pulls out variation orthogonal to fixed country factors NGM performs very poorly “Combining these growth results with our findings on levels, we call for returning productivity differences to the center of theorizing about international differences in output per worker.”

82 Hsieh (2002) Intuition Young: Tigers grew through rapid capital accumulation with modest TFP growth If true, market return to capital should have fallen Can construct alternative estimates of capital accumulation using rental rates (prices) rather than national accounts (quantities) – the “dual” method

83 Methodology Both estimates of the Solow Residual should coincide if accounting data are accurate If they differ, we may trust prices more than government accounts

84

85 Synopsis Convergence/divergence Levels/growth accountingOverall divergence Conditional convergence suggests ample scope for poor countries to benefit from the rich Through what channels? Levels/growth accounting Measurable capital (physical, human) does matter Lots of space in the residual for technology, poverty traps, institutions