The Employment of Nations Introduction • Material living standards depend on the availability of goods and services • Availability depends on production; production depends on employment (more precisely: the aggregate labor input) • What determines how much time (a scarce resource) is allocated to production in an economy? • In short: what determines the employment of nations? The Aggregate Labor Input • Extensive Margin: Number of people employed (employment) • Intensive Margin: Hours worked per employed person • A commonly-used measure of the aggregate labor input: Employment Hours Worked per Employed Person Hours per capita = Adult × Population Available Time per Employed Person Table 2.1 Aggregate Employment Rates Selected OECD Countries Canada United States Japan Australia Denmark France Germany Ireland Italy Netherlands Spain Sweden United Kingdom 1965 1970 1975 1980 1985 1990 1995 60.9 61.3 63.4 65.5 66.1 68.9 67.3 62.9 64.5 64.4 66.8 68.5 72.4 73.4 71.5 71.1 69.7 70.3 70.6 72.5 74.4 66.6 68.8 67.5 65.5 64.7 67.6 67.4 72.2 74.3 73.3 73.1 75.3 76.8 74.1 66.4 65.8 65.4 63.5 59.4 60.1 58.9 69.6 68.6 65.7 63.6 62.5 65.8 65.2 64.4 62.0 58.0 57.0 51.7 52.7 55.2 58.8 56.3 55.6 55.7 54.0 55.0 52.5 59.9 57.3 54.5 53.6 53.3 61.5 65.3 59.8 60.1 58.1 50.2 45.4 48.9 46.4 71.9 72.7 76.3 78.7 79.3 79.8 70.4 71.4 70.4 70.8 68.7 65.7 70.7 68.6 Mean Standard Deviation 65.7 5.71 65.7 5.91 65.4 6.38 64.8 7.67 63.8 9.45 66.7 9.30 65.9 8.82 Table 2.2 Annual Hours Worked per Employed Person Selected OECD Countries Canada United States Japan France Germany Italy Norway Spain Sweden 1970 1890 1889 2201 1962 1949 1969 1766 n.a. 1641 1975 1837 1832 2112 1865 1801 1841 1653 n.a. 1516 1979 1832 1845 2126 1806 1696 1722 1514 2022 1516 1983 1780 1808 2095 1712 1657 1699 1485 1912 1518 1990 1788 1819 2031 1657 1598 1674 1432 1824 1546 1996 1784 1839 1892 1608 1511 1636 1407 1810 1623 Mean 1908 1807 1757 1719 1693 1663 Standard Deviation 163.1 172.2 197.8 189.9 185.0 165.6 Figure 2.1 Hours Worked as Fraction of Total Time Selected OECD Countries 0.25 0.2 0.15 0.1 0.05 0 1965 1970 1975 1980 1985 1990 1995 2000 Approach • Observation: large and persistent differences in employment patterns across nations • Question: what accounts for these differences? • Aggregate behavior is simply the sum of individual behaviors • Implication: to understand aggregate behavior, we need to understand what motivates the individual behavior that generates it • We need a theory of individual decision-making, applicable to the question at hand • Let us begin with two self-evident facts: 1. People are endowed with a limited amount of time; 2. People generally have competing uses for this time. • Implication: people must decide how to allocate scarce time across competing uses (this is called a time-allocation problem) • Approach: build a model economy, populated by model people, similar (but by no means identical) to real people along a few selected dimensions • We need to address a few questions: 1. What motivates our model people? (Specify an objective) 2. What actions are under their control? (Endogenous variables) 3. What objects are beyond their control? (Exogenous variables) 4. How do objectives interact with constraints to determine behavior? (Prediction) A Basic Model Economy • There are two commodities: consumer goods () and home goods () • The pair ( ) ∈ R2+ is called a commodity bundle (where R2+ constitutes the commodity space) • What motivates a model person? Assume that they derive pleasure from consuming ( ) • Moreover, assume that they can attach a number ( ) to each ( ) ∈ R2+ that ranks each commodity bundle in the commodity space • Assume that ( ) is strictly increasing in and ; and strictly concave in and 0 and 0 2 2 0 and 0 2 2 • Common example: ( ) = ln() + ln(); ≥ 0 • An indifference set is defined to be all combinations of ( ) that are equally ranked; i.e., that satisfy ( ) = 0 for some 0 ∈ R (note: 0 could be negative) • Note: given the assumptions made on this indifference set will be an “indifference curve” that slopes downward in a convex manner from left to right in the commodity space • The marginal rate of substitution or MRS is (the absolute value of) the slope of the indifference curve • Denote by ( ); measures the relative valuation attached to and by an individual at the point ( ) • Also: measures an individual’s willingness to substitute across different commodities • Additional assumption: homotheticity Example ( ) = ln() + ln() • Now set ( ) = 0 so that ln() + ln() = 0 • This implies an indifference curve = 0 − • Slope is • So MRS is given by µ ¶ = −0 −−1 = − µ ¶ ( ) ≡ • Note: these preferences are homothetic Constraints • Let denote time endowment • Let denote skill level in producing consumer goods • Let denote time devoted to producing consumer goods (employment) • Let denote time devoted to producing home goods (leisure) • Assume that = (can only consume home goods produced by self) • Then constraints are given by + = = • Or, combining... = − • Think of this as a budget constraint; where denotes the price of measured in units of (the real wage) • Alternatively + = ; where denotes the value of your time endowment measured in units of Summing Up • A person has preferences defined over ( ) ranked by a function ( ) • Assume that ( ) are choice variables • Therefore, choose the ( ) that delivers the highest rank ( ) • But without violating the budget constraint ≤ − • In short, the choice problem can be stated as ( ) ≡ max { ( ) : ≤ − } ( ) ≡ max { ( ) : ≤ − } • The choice variables ( ) are endogenous variables (objects that the theory is designed to predict/explain) • The parameters ( ) are exogenous variables (objects that are not explained, but which potentially influence behavior) • The solution to the choice problem is an optimal ( ); this solution will be a function of parameters ( ) • ( ) is a maximum value function; or indirect utility function (think of it as a welfare function) Review • Model economy consists of people with preferences defined over commodity bundles ( ) • Preferences are represented by a ranking (or utility) function ( ); e.g., ( ) = ln() + ln() • ( ) measures a person’s willingess to substitute across and • Model people endowed with time and skill • There are two uses of time: + = (time constraint) • Assume that = ; so that ≤ • Constraint boils down to ≤ − • measures a person’s ability to substitute across and (interpret as real wage; it is the price of leisure measured in units of consumption) • Now, assume that people try to do the best they can, subject to their constraints • In this model, choose ( ) to maximize ( ) subject to ≤ − • Solution is a pair of functions ∗( ) and ∗( ); which implies ∗( ) = − ∗( ) • Note: solution is a conditional forecast • That is, this theory has the structure ( ) ⇒ (∗ ∗ ∗) • We can use the theory to predict and interpret how any exogenous change in ( ) will affect individual behavior • We can also use theory to evaluate welfare consequences; ( ) ≡ (∗ ∗) • Mathematically, solution is characterized by (∗ ∗) = ∗ = − ∗ • Example 1: ( ) = ln() + ln() implies ( ) = (); so solution is µ ¶ 1 ∗ = 1 + ¶ µ ∗ = 1 + ¶ µ 1 ∗ = 1+ • Note: ∗ is increasing in ; while time allocation is invariant to (interpretation: SE and WE stemming from a change in exactly cancel) • Example 2: ( ) = 212 + 212 implies ( ) = ()12; so solution is ∗ = à 2 ! 1 + 2 à ! 1 ∗ = 2 1 + à ! 2 ∗ = 2 1 + • Note: ∗ is increasing in ; and now ∗ is increasing in (so that ∗ is decreasing in ) • For these preferences, SE dominates WE From Individual to Aggregate Behavior • Let Π(X ) denote fraction of population with characteristics ( ); Π( ) = 1 so that • Special Case [Representative Agent Model] is when Π( ) = 1 for some ( ) • Aggregate labor input is given by ∗ = X ∗( )Π( ) • For representative agent model, aggregate labor input is simply ∗( ) Interpreting Cross-Country Differences in Employment • Fact: over long periods of time, aggregate labor input (per capita) is approximately constant (in relatively advanced economies) • One interpretation: for permanent wage changes, SE and WE roughly cancel • This suggests that cross-country wage/productivity differentials is not likely the primary source of observed differences in employment patterns • Another possibility is cross-country differences in ; but not generally considered a plausible explanation Redistributive Policies • Ed Prescott argues that cross-country differences in tax/transfer policies may be important • In fact, there appear to be large and persistent differences in tax rates across countries Figure 2.11 Average Tax Rate on Labor Income Selected OECD Countries 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 1965 1970 1975 1980 1985 1990 1995 2000 • Simple explanation is that higher labor income tax reduces the return to working; leading to a substitution away from work into leisure • In fact, the argument is a little more subtle than this • In particular, a higher labor income tax is like a reduction in the real wage • But if SE and WE roughly cancel, not likely to have a big impact on time allocation choices • Key lies here: what does the government do with tax revenue? • Assume that they redistribute it in lump-sum manner (not a bad approximation) Modeling a Redistribution Policy • Let 0 ≤ ≤ 1 denote labor income tax rate; and let ≥ 0 denote lump-sum transfer • Then an individual’s budget constraint is ≤ (1 − )( − ) + • The slope of the budget line is now −(1 − ) • That is, an increase in makes the budget line “flatter” (distorts the relative price of consumption and leisure) • So an increase in increases the price of consumption relative to leisure (leisure is now relatively cheaper) • On the other hand, an increase in is like an increase in wealth • Choice problem is max { ( ) : ≤ (1 − )( − ) + } • Solution is characterized by (̂ ̂) = (1 − ) ̂ = (1 − )( − ̂) + • Assume that people only differ in ; then solution is a pair of functions ̂( ) and ̂( ) • Example: ( ) = () µ ¶ µ ¶µ ¶ 1 ̂ = − + 1+ 1 + à 1 − ! µ ¶ µ ¶ + ̂ = 1+ 1+ (1 − ) à ! µ ¶ µ ¶ 1 ̂ = − 1+ 1+ (1 − ) • Notice: reduces to what we had before if = = 0 • Theory predicts that increase in generally — decreases increases decreases • and that increase in generally — increases increases decreases • So an increase in both ( ) appears to have ambiguous impact on ; but clearly leads to a decrease in • Think about it this way: — increase in implies ↓ (SE) and ↑ (negative WE) — increase in implies ↑ (positive WE) • Negative and positive WE basically cancel; leaving the SE • But why should policy increase both and ? Why not just set tax rate to zero and give everyone a transfer? • Answer: Because such a policy is not budget feasible for the government The Government Budget Constraint (GBC) • In this model economy, the GBC is given by = X ̂( )Π() • So, for exogenous policy this can be solved for the budget-balancing tax rate ̂ • Or, for exogenous policy this can be solved for the budget-balancing lump-sum transfer ̂ A Representative Agent Economy • For a given theory delivers the following restrictions (̂ ̂) = (1 − ) ̂ = (1 − )( − ̂) + ̂ ̂ = ( − ̂) (1) (2) (3) • Hence, 3 equations in the 3 unknowns (̂ ̂ ̂ ) as a function of parameters (in particular, ) • And, of course, once we know ̂ we can deduce ̂ • Condition (2) asserts that (̂ ̂) lies on the budget line ̂ = (1 − )( − ̂) + ̂ • If we combine (2) and (3), we get ̂ = ( − ̂) — this implies that (̂ ̂) also lies on the feasible line • Now, since (̂ ̂) = (1 − ) = (∗ ∗) it follows that ̂ ∗ and ̂ ∗ for any 0 • It follows that ̂ is decreasing in ; which is broadly consistent with the data • For a RA economy, economic welfare is also decreasing in A Heterogeneous Agent Economy • Assume that people differ according to only; distributed according to Π() • Now, for any given we can compute an equilibrium allocation ̂( ) ̂( ) for each — we can also compute individual welfare ( ) ≡ (̂( ) ̂( )) • Generally speaking, low types welfare is increasing in ; while high types welfare is decreasing in (redistribution) • The redistribution policy modeled here is called a negative income tax (NIT) • People pay tax and receive transfer ; so that their net tax obligation is − • For high types, − 0 (net contributors) • For low types, − 0 (net recipients) • Increasing (and hence ) has the effect of increasing inequality in gross earnings • Why? • Because ↑ leads low-skill to reduce labor supply more than high-skill • Why? • Because high-skill are net contributors (negative wealth effect implies greater willingness to work) • Because low-skill are net recipients (positive wealth effect implies they can afford more leisure) • On the other hand, increasing (and hence ) has the effect of reducing inequality in after-tax earnings (1 − ) + Socially Optimal Tax Policy • The answer to this question depends on what one assumes is a reasonable measure of social welfare • A case could be made for the Rawlsian veil of ignorance W= X ( )Π() • If government objective is to maximize W then generally speaking, optimal tax rate is positive • Essentially, redistributive program insures risk-averse agents against the “bad luck” of being born with low skill Taking the Theory to the Data • Assume ( ) = ln() + ln() and identical across people and countries (representative agent model) • This implies ∙ 1− ̂( ) = +1− ¸ • Calibrate this to the U.S. in 1985; where = 01414 and = 0209 • This implies ∙ 1 − 0209 01414 = + 1 − 0209 ¸ • Solve for = 4803 • Now, we can use our “estimated” model’s prediction ∙ 1− ̂( ) = 4803 + 1 − ¸ • Data is for Canada, France, Germany, Italy, Japan, Spain and Sweden; each in the years 1970 and 1995 Figure 2.14 Actual and Predicted Hours Worked Hours Worked / Total Time 0.2 0.18 0.16 0.14 0.12 Predicted 0.1 0.08 0.06 Actual 0.04 0.02 0 0 0.1 0.2 0.3 Tax Rate 0.4 0.5 Conclusions • In the context of the theory developed here, cross-country differences in employment not likely the consequence of cross-country differences in productivity • Cross-country differences in the “generosity” of redistributive programs may be quantitatively important (but clearly not the entire story) • Higher labor-income tax rates may induce lower levels of employment and per capita income (GDP), but this does not necessarily imply that lower GDP is associated with lower “social” welfare ⇒ do not confuse GDP or employment with social welfare