The Supply Side of the Economy

TOPIC 2 The Supply Side of the Economy

Goals of Lecture 2 Introduce the supply side of the macro economy. Discuss how countries grow and why some countries grow faster than others. Discuss labor productivity o What does it mean? o How does it respond coming out of recessions? Determine how wages are set in an economy. Determine why people work. Understand where unemployment comes from. 2

The Production Function GDP (Y) is produced with capital (K, price-weighted) and labor (N, hours): Y = A F(K,N) Sometimes, I will modify the production function such that: Y = A F(K,N, other inputs) where other inputs include energy/oil! Realistic Example is a Cobb Douglas function for F(.): Y = A K 1-α N α A is Total Factor Productivity (TFP), an index of efficiency (technology) MUST READ: NOTES 3 (my text posted on the teaching page) on the aggregate production function 3

Measurement Y is GDP (measured in dollars). As noted above, we want to measure Y in real dollars.<<you should know what this means from Notes 1 of the text>>. For our Cobb Douglas production function (previous slide), N and K are both measured in dollars. N often is measured in total wage bill K often is measured as the replacement cost of capital However, in practice, N can be measured in different ways (hours worked, number of workers). Wage bill is the preferred method (takes into account skill differentials). However, we will often talk about standard of living which is income per capita (Y/N ; where Y is income and N is some population measure). 4

Features of the Aggregate Production Function Define MPN = Marginal Product of Labor = dy/dn Define MPK = Marginal Product of Capital = dy/dk Math Note: You should be comfortable taking these simple partial derivatives if you are not, practice this for the quizzes and exams. Diminishing Marginal Products From Cobb-Douglas: MPN = α A (K/N) 1-α = α (Y/N) Fixing A and K, MPN falls when N increases MPK = (1-α) A (N/K) α = (1-α) (Y/K) Fixing A and N, MPK falls when K increases Complementarity Across Inputs Increasing A or K, increases MPN Increasing A or N, increases MPK 5

Labor Share With Cobb-Douglas Labor Share of Income = Income earned by workers divided by GDP = [(W/P)*N/]Y In equilibrium, real wages of workers will equal MPN (more on this below) Substituting (W/P) = MPN into above yields: Labor Share of Income = (α) * (Y/N) * (N/Y) (MPN) = α Cobb-Douglas predicts a constant labor share of α. Historically, α was stable at a level of about 0.7 (in notes, I often just set α = 0.7). 6

US Labor Share: 1947Q1 2016Q2 Sharp Decline After 2000 Roughly Stable through 2000 7

Manufacturing Labor Share : 1988Q1 2014Q2 8

Sub-Section A Economic Growth

Two Measures of Productivity Labor Productivity = Y/N = A (K/N) 1-α Driven by A and K/N (usually reported in press) Total Factor Productivity (TFP) = A = Y/F(K,N) Basically TFP is a catch-all for anything that affects output other than K and N. Work week of labor and capital Quality of labor and capital Regulation Infrastructure Competition Specialization Innovation (including innovation in management practices) Changes in discrimination or culture Some components of TFP tend to be pro-cyclical (Definition of Pro-cyclical: Variable increases when Y is high, decreases when Y is low) 10

Growth Accounting Y = A K 1-α N α (our production function) %ΔY = %ΔA + (1-α)%ΔK + α%δn Output, in a country grows from: Growth in TFP (see entrepreneurial ability, education, roads, technology, etc.) Growth in Capital (machines, equipment, plants) Growth in Hours (workforce, population, labor participation, etc). Perhaps, we care about growth in Y/pop or Y/N (per capita output). %Δ(Y/pop) = %ΔA + (1-α)%Δ(K/pop) + α%δ(n/pop) or %Δ(Y/N) = %ΔA + (1-α)%Δ(K/N) 11

How is TFP Measured The way TFP (A) is usually measured is via a statistical decomposition (referred to as the Solow Residual ). Remember our assumed production function: Y = AK 1-α N α Math Note: We are going to transform the production function to make it a little easier to work with (you should get comfortable with this) by taking the logs: ln(y) = ln(a) + α 0 ln(k) + α 1 ln(n) (where α 0 = 1 α 1 0.3) (1) Given that we measure Y, K and N in the data, we can estimate (1) using standard regression techniques. ln(a) is the constant from the regression. This is our standard TFP measure. 12

US TFP Growth: 1970Q1 2014Q1 13

Measuring TFP Because A (TFP) is a catch-all term for anything that affects production, the assumed production function does not impose any structure on how to measure the components of TFP. Economists are very good at measuring the extent to which TFP changes over time within a country. It is much harder to measure why TFP has changed over time. Economists try to measure this by using detailed firm-level and household-level data to measure production and wages. 14

How is Labor Productivity Measured Labor Productivity is easier to measure: Y/N Y is usually real GPD N is usually total hours worked 15

Nonfarm Business Labor Productivity Growth: 1988Q1 2017Q1 Puzzle? 16

Manufacturing Labor Productivity Growth: 1988Q1 2016Q2 Puzzle? 17

U.S. Labor Productivity Growth Over Time 1900-1972: 2.4% per year 1972-1996: 1.4% per year. 1996-2004: 2.6% per year. ( Internet boom ) 2005-2016: 1.1% per year 2013-2016: ~0% per year Data from Robert Gordon (Professor at Northwestern) 18

Sub-Section A1 The Causes of Economic Growth

The Role of Investment and Growth Does a one time increase in investment today increase Y/N today? YES! Does a one time increase in investment today cause a sustained increase in Y/N into the future? No! Back of our mind equations: S = I + NX (From the first lecture). K(t+1) = (1-δ) K(t) + I(t) or ΔK(t, t+1) = I(t) - δ K(t) Notice the link between saving and investment. Definition of Capital Stock Evolution All else equal (i.e. holding N constant), increasing I causes K tomorrow to increase causing K/N tomorrow to increase (i.e. Y/N tomorrow increases). 20

Time Path of Capital Stock: One Time Increase in I K No investment No investment t+1 time Suppose there is a one time increase in investment at time t (perhaps due to an investment tax credit). Suppose no investment either prior to or after the tax credit. 21

Can Higher Investment Lead to Infinite Growth? Does a sustained increase in investment increase Y/N today? YES! Does a sustained increase in investment cause a sustained increase in Y/N? No! Suppose I is fixed at a high level and that K initially is sufficiently small. K grows if I > δ K: But, notice that δk is also growing each period. (Summary: To start, higher I will lead to higher K and Y/N will increase). Eventually, however, I will converge towards δk. More and more of the investment is going to replace outdated capital and the capital stock will grow by smaller and smaller rates. The increase in Y/N will converge back to zero. Summary: High levels of investment will increase the capital stock and output, but both K and Y will eventually converge to a fixed level. 22

Time Path of K: Permanent Increase in I K The new level of investment has successively less effect due to growing depreciation of the capital stock. No investment t+1 time Suppose there is a permanent increase in investment at time t. Suppose no investment prior to t. In all periods after t, the level of investment remains fixed at the level in t. 23

Can Higher Investment Growth Cause Infinite Growth? If a one time increase in I gives an increase in Y, why not continuously raise I to higher and higher amounts??? Answer: Diminishing MPK!!! MPK =.3 A (N/K).7 ; As K increases, MPK falls. As K goes to infinity, MPK goes to zero (Y stops increasing). Suppose, we keep rising I (each year), K will increase by the amount of I (after controlling for depreciation), but Y will increase by continuously smaller and smaller amounts. Remember Y = C + I + G + NX. I/Y (investment rate) is bounded by 1 (if you invest all your output). This caps the increase in I. I cannot grow forever! Continuously increasing I will NOT lead to sustained economic growth. NOTE: Investment decisions are NOT made in the dark (i.e. something must drive firm investment!) 24

What Causes Sustained Growth? Sustained Increases in the growth of A are the only thing that can cause a sustained growth in Y/N. Empirically, when a country exhibits faster Y/N growth.. 33% typically comes from growth in K/N 67% typically comes from growth in A (where N = employment (not hours) - limited data). 25

Nonfarm Business Labor Productivity Growth: 1988Q1 2017Q1 Puzzle? 26

Manufacturing Labor Productivity Growth: 1988Q1 2016Q2 Puzzle? 27

Why is US Labor Productivity/TFP Growth Lower Demographics? Slower growth in schooling? Innovations with less spill-overs on productivity (but huge spill-overs on leisure) What productivity growth we have seen replaces lower skilled workers. Contributing to inequality within the economy. Have we moved from big innovations in market productivity (electricity, assembly lines, transportation, manufacturing techniques, etc.) to big innovations in leisure technology (facebook, internet, games, etc.)? Less competition/more regulation? 28

New Paper of Mine The Allocation of Talent and Economic Growth Question 1: How much of the observed TFP growth in the U.S. since 1960 is due to better labor market outcomes (including human capital formation) for blacks and women? o o A better allocation of resources leads to higher economic growth. There have been large changes in the allocation of women and minorities to occupations in the labor market since 1960. Question 2: How much of the convergence of the U.S. south to the U.S. north is due to a decline in discrimination of the south? 29

Occupational Sorting Over Time: An Overview Fraction of group (white men, white women, black men, black women) aged 25-55 working in the following occupations: Executives, Mgmt, Architects, Engineers, Math/Computer Science, Natural Scientists, Doctors, and Lawyers. 1960 2010 White Men 21.2% 23.5% White Women 3.0 (7.3) 17.4 (21.0) Black Men 2.8 14.6 Black Women 1.0 (2.1) 13.0 (15.2) Data: U.S. Census and American Community Survey

Occupational Sorting Over Time: An Overview Where were the other groups working in 1960? 53% of working white women worked in Nursing, Teaching, Sales, Secretarial and Office Assistances, and Food Prep/Service. o The comparable number for white men was 14% (mostly sales) 55% of working black men worked as Freight/Stock Handlers, Motor Vehicle Operators, Machine Operators, Janitorial Services, and Personal Services. o The comparable number for white men was 19% 47% of working black women worked in Household Services, Personal Services, and Food Prep/Services. o The comparable number for white men was 2%

Wage Gaps Over Time: An Overview Log difference in annual earnings of full time workers, conditional on experience, hours and occupation controls (relative to White Men) 1960 1980 2010 White Women -0.56-0.47-0.26 Black Men -0.37-0.21-0.16 Black Women -0.82-0.47-0.31

Findings Macro Implications: o 25% 30% of per capita earnings in the U.S. between 1960 and 2010 was due to declining frictions for white women, black women, and black men. (Shines some light into the black box of TFP growth) o Other interesting results: - Wage growth in the 1970s and the 2000s would have been negative absent the labor market improvements for blacks and women. - About 40% of the convergence of the south to the northeast between 1960 and 1980 is due to declining labor market frictions. - Not much remaining room for growth from this mechanism. Can explain some of the slow down in productivity since 2000.

Another (Potentially Crazy) Idea I have on Productivity Slowdown

Did Early 2000 Finance Boom Cause a Fall in STEM Employment?

Is 3 or 4 percent growth possible? Some politicians promise growth rates of 3 or 4 percent per year? o o They do not specify but, lets suppose they mean real growth. They do not specify growth in Y or Y/N the former is possible by increasing N. Is 3 or 4 percent growth per year possible? o o o Possible maybe for short periods of time. However, given U.S. history, I do not think it is likely. Also, given current economic conditions, I do not think it is likely at all. Can Presidents and Congress affect short run growth rates? o o No! (unless we are in a recession) At best, policy makers can affect growth rates in the long run by affecting TPF. 36

Summary There is a productivity slowdown likely a combination of many factors. Do not believe politicians who tell you they can magically raise growth in Y/N to 3 or 4 percent per year. That is crazy talk. I would be ecstatic if we could increase growth to about 2% per year (i.e., return to our long run average). Growth in immigration (N) can increase Y but not necessarily Y/N unless it affects A. Note: Demand side factors do not affect growth conditional on A, K and N. Conditional on A, K and N, demand side factors will only affect prices. (We will prove that later in the course). 37

Sub-Section A2 Cross-Country Growth

Growth Across Countries Most developed economies grow at the same rate that the technological frontier grows. Some helpful definitions: Convergence countries inside of the technological frontier move towards the technological frontier. Divergence countries inside of the technological frontier grow at a rate less than the technological frontier. 39

Distribution of World GDP in 2015 (IMF, $) 40

World GDP Per Capita in 2016 (IMF, $) Top 10 Other Notable Bottom 10 Qatar 127,660 Lithuania 29,972 Madagascar 1,505 Luxembourg 104,003 Portugal 28,933 Eritrea 1,410 Singapore 90,151 Russia 26,490 Guinea 1,265 Brunei 76,884 Argentina 20,047 Mozambique 1,215 Kuwait 71,887 Iraq 17,944 Malawi 1,134 Norway 69,249 China 15,399 Niger 1,107 Ireland 69,231 Brazil 15,242 Liberia 855 U.A.E 67,871 South Africa 13,225 Burundi 814 Switzerland 59,561 Ukraine 8,305 Congo, Dem. Rep. 773 USA 57,436 India 6,616 Cent Africa Rep. 652 Note: Purchasing price parity adjusted 41

Some Data: Distribution of World GDP in 2000 From Barro, 2003 includes 147 countries. Horizontal axis is a log scale. All data are in 1995 U.S. dollars. 42

Some Data: Distribution of World GDP in 1960 From Barro, 2003 includes 113 countries. Horizontal axis is a log scale. All data are in 1995 U.S. dollars. 43

Growth Rate of GDP Per Capita: 1960-2000 From Barro, 2003 includes 111 countries. 44

GDP per Capita in the United States, the United Kingdom, and Japan, 1870 2009 (Weil 2015) Sources: Maddison (1995), Heston, Summers, And Aten (2011).

The Distribution of Growth Rates, 1975 2009 (Weil 2015)

Convergence of Income Across U.S. States: 1940-1980 Historical Trends in Convergence Unadjusted 1940-1960 Growth in Per Capita Income 1940-1960.2.4.6.8 1 MS ARAL ND SD KY OK NC GA NM KS NE TN SC LA TX UT CO ID IA MO WV VA AZMNWI IN FL WY VT NH ME MT OR WA OH PA MI MD IL RI MA CA NJ NYNV CT DE 2000 4000 6000 8000 10000 12000 Per Capita Income 1940 Fitted values gr_ipc_40_60 47

Convergence of Income Across U.S. States: 1980-2000 Recent Trends in Convergence Unadjusted 1980-2000 Growth in Per Capita Income 1980-2000.1.2.3.4.5 MS AR NC GA SCND SD VT TN ME AL KY WV NM UTID LA MT NE MO IN IA AZ OK RI NH MN VA PA TX FL MI KS OH WI OR MA CO NY WAIL DE MD NJ CA NV WY CT 15000 20000 25000 Per Capita Income 1980 Fitted values gr_ipc_80_00 48

Source of GDP Growth Latin America Brazil, Chile, Columbia, Mexico, Peru, Uruguay, Bolivia, Ecuador, Paraguay, Venezuela Emerging Asia Indonesia, Malaysia, Philippines, Thailand, and China Advanced Exporters Australia, Canada, New Zealand, and Norway. From Sosa et al. (2013), IMF Report 49

Per Capita GDP vs. Life Expectancy (Acemoglu 2005) 50

Sub-Section B The Labor Market

Labor Market: Firm Profit Decisions In a competitive market, a firm can sell as much Y as it wants at the going price p, and can hire as much N as it wants at the going wage w. Facing w and p, a profit maximizing firm will hire N to the point were MPN = w/p (the benefit from an additional worker (in terms of additional output) must equal the cost which they are paid). <<This is straight from micro>> With Cobb-Douglas: MPN =.7 Y/N =.7 A (K/N).3 If firms maximize profits: w/p =.7 Y/N =.7 A (K/N).3 If MPN > w/p then the firm can increase profits by increasing N. If MPN < w/p then the firm can increase profits by decreasing N. Reading: Notes 4 from the supplemental notes 52

The Labor Demand Curve w/p * real wage N* MPN = N d N 53

Notes on the Labor Demand Curve N d slopes downward (N d = MPN =.7 A * (K/N).3 ) N d rises with A and K (assumed complementarity across inputs) Assumption: Y is not Fixed! Firms optimally choose N, K, Y and (to some extent) A to maximize profits. Caveat: Who says that there is a demand for more Y? Need to look at the demand side of economy (introduced last -discussed in depth throughout the course). 54

The Other 1/2 of the Labor Market: Labor Supply Labor Supply (N s ) Results from Individual Optimization Decisions Households compare benefits of working (additional lifetime resources) with cost of working (forgone leisure) Factors Affecting Labor Supply The Real Wage (w/p) The Household s Present Value of Lifetime Resources (PVLR) The Marginal Tax Rate on Labor Income (t n ) The Marginal Tax Rate on Consumption (t c ) Value of Leisure (reservation wage) - non- work status (VL) The Working Age Population (pop) 55

The Labor Supply Curve N s (PVLR, t c, t n, pop, VL) w/p N 56

Labor Supply Notes (Most Derived From Scratch in Lecture) In terms of wages and earnings, there is both an income and substitution effect - we will look at them separately BUT in the real world, they often occur jointly!!!! The Real Wage - HOLDING PVLR fixed: A higher w/p encourages individuals to substitute away from leisure and toward work (leisure becomes more expensive). This is a substitution effect. <<This is why the labor supply curve slopes upwards!!>> Estimating this substitution effect is difficult since PVLR is not easily held constant. Estimates range from 0-2% (For a 1% increase in after-tax w/p holding PVLR fixed, labor supply either increases by 0% or 2%). Very Wide Range little consensus. PVLR = initial wealth + present discounted value of earnings A higher PVLR induces individuals to work less (lower N s ) for a given after-tax wage, allowing them to enjoy more leisure (If leisure is preferred to work as I get richer, I can afford to work less). PVLR is net of taxes and non-work governmental transfers and inclusive of all other transfers. 57

Labor Supply Notes Marginal tax rate on labor income - Should have same substitution effect as the before tax real wage. Studies of the 1986 U.S. Tax Reform found that only high-earning married women worked more in response to lower marginal income tax rates. Marginal tax rate on consumption - see above Value of Leisure - If leisure/no-work becomes more/less attractive, households will work less/more (reservation wage). (Welfare programs, child care, etc.). Working Age Population: Usually defined as 16-64. (Includes changes in Labor Force Participation Rates) 58

Recap on Labor Supply Substitution Effect: For a given PVLR, a higher after tax wage increases N S. Income Effect (This is why Labor Supply Curve Slopes Upward) For a given after-tax wage, higher PVLR decreases N s. Evidence: Weak Consensus is that, with equal (%) increase in PVLR and the after-tax wage, N s falls (income effect dominates). 59

Temporary Increase in A N s (PVLR, t c, t n, pop, VL) w/p w/p * N* N d (A,K) N 60

Permanent Increase in A N s (PVLR, t c, t n, pop, VL) w/p w/p * N* N d (A,K) N 61

Can Technological Progress Destroy jobs? Facts: A, N, w/p are trending up over time. N/pop is trending down (except in U.S. since 1980). Higher A countries have higher w/p and lower N/pop. Implications: Adjusting for pop, higher A goes with lower N. Higher A reduces N d and destroys jobs? - NO! Labor Demand Increases. Higher A increases PVLR and reduces N s - The Effect on Labor Supply is to fall. 62

Permanent Increase in Population N s (PVLR, t c, t n, pop, VL) w/p w/p * N* N d (A,K) N 63

Population and Jobs More People = More Jobs 1000000 1990 Employment (000s) 100000 10000 1000 100 100 1000 10000 100000 1000000 1990 Working-Age Population (000s) 64

Temporary Increase in Taxes (t c or t n ) N s (PVLR, t c, t n, pop, VL) w/p w/p * N* N d (A,K) N 65

Permanent Increase in Taxes (t c or t n ) N s (PVLR, t c, t n, pop, VL) w/p w/p * N* N d (A,K) N 66

Labor Market Equilibrium (in long run!) We define Long Run Equilibrium in macroeconomics as occurring when the labor market clears. By definition, long run macro equilibrium exists when N = N*. At N*, labor demand = labor supply. So, by definition, all workers who want a job (the suppliers) are able to find a firm looking for a worker (the demanders). Implies that cyclical unemployment = zero at N*. Long run equilibrium is characterized by zero cyclical unemployment! It is an equilibrium in that there is no incentive for real wages to change at N* Real wages (w/p) has two components: nominal wages (w) and the price level (p). Note: Y* (by definition) = A K.3 (N*).7 Y* is the long run equilibrium level of output (output where labor market is in equilibrium) 67

Our First Aggregate Supply Curve. Suppose prices (p) increase. What happens in the labor market? In terms of equilibrium, nothing happens! Increasing prices have no effect on labor demand (A and K do not change). Increasing prices have no effect on labor supply (taxes, population, etc. do not change). You may ask Doesn t PVLR change when prices increase??? No! As long as nominal wages adjust, real wages will be unchanged when p increases. The % change in prices will be match exactly by the % change in nominal wages real wages will not change (so PVLR will not change). No effect on labor supply. Key: Because real wages will not change, changes in prices will have NO effect on the labor market (i.e., it will have no effect on N*). Conclusion: Changing prices will have NO effect on Y* (since N* is constant). 68

Our First Aggregate Supply Curve p LRAS Long Run Aggregate Supply Curve Y* Y If labor market clears, changes in prices will lead to equal changes in nominal wages. As a result, there will be no change in N* and hence, no change in Y*. Leads to a vertical LRAS curve. Prices do not affect production in the long run! 69

What Shifts Y*? (the LRAS) Anything that affects the labor market will affect Y*! If N* increases, Y* will shift to the right. If N* decreases, Y* will shift to the left. Summary: Y* will shift right if: A increases K increases population increases labor income taxes fall (and income effect is small relative to substitution effect) labor income taxes rise (and income effect is large relative to substitution effect) Note: The long run aggregate supply curve (LRAS) is NOT the labor supply curve. We have lots of different markets in this class. There will be lots of different supply and demand curves. You need to keep track of them! 70

Things to Remember! The demand side of the economy is NOT important for determining Y*! All we need to know is A, K and N and we know Y*! The demand side of the economy is not important for economic growth! Key: If I ever ask you about what determines Y* (i.e., output/income/expenditure in the long run), you should think about A, K and the labor market. As a rule, K will be fixed unless I tell you otherwise (for simplicity, you will see why soon). Why do we care about the demand side of the economy? In the long run, prices will be determined by demand. Also, LRAS is dependent on labor market being in equilibrium. In the short run, labor market need not be in equilibrium. Demand will determine output in the SHORT RUN! 71

Summary. In the long run when labor markets clear. Supply side of economy (labor market, K, A, other inputs like oil) determines output. Demand side of economy (C+I+G+NX) will determine prices. In the short run when labor markets do not clear: Demand and Supply jointly determine prices and output (think of the simple examples I gave graphically in the lecture for topic 1). Three outstanding issues (we will get to them soon): When is the labor market NOT in equilibrium? What does the supply curve look like when labor market doesn t clear? What determines demand? 72

When are Labor Markets in Disequilibrium? Labor market is in disequilibrium when labor demand is not equal to labor supply. Any time labor demand = labor supply, there is no cyclical unemployment (by definition)! Nominal wages do not adjust to clear the labor market We refer to this as sticky wages. Because of wage contracts (and uncertainty), nominal wages do not always adjust immediately. Need a model for short-run disequilibrium --- we will do that in Topic 6. 73

Distribution of 12 month Wage Change, Stayers Hourly Salaried Percent 0 10 20 30 40-50 -40-30 -20-10 0 10 20 30 40 50 Wage Change (%, 12-month) Percent 0 10 20 30 40-50 -40-30 -20-10 0 10 20 30 40 50 Wage Change (%, 12-month) Note: Large mass at zero 35% of hourly and 33% of salaried unchanged Note: Hardly any wage cuts 1.5% of hourly and 3% of salaried

Distribution of 12 month Wage Change, Excluding Zeros 2009 Post-2010 Percent 0 10 20 30-50 -40-30 -20-10 0 10 20 30 40 50 Wage Change (%, 12-month) Percent 0 10 20 30-50 -40-30 -20-10 0 10 20 30 40 50 Wage Change (%, 12-month) Wage cuts more likely to occur during the Great Recession 5.5% overall, 9.1% for salaried in 2009 vs 2.1% and 3.0% in 2011-2013.

Cyclical Unemployment in Labor Markets When do we get cyclical unemployment in our models? Cyclical unemployment occurs when there are no jobs available (labor demand) for those with the skills and the desire to work (labor supply) at current wages. Cyclical unemployment occurs only in disequilibrium! (when desired labor demand < desired labor supply - at given wages) N s w /p b a Unemployment N d N(1) N(0) 76