Interpreting the Business Cycle Business Cycles • Real per capita GDP does not evolve in a “smooth” manner over time — typically fluctuates around “trend” — sometimes, “trend” appears to shift as well • Fluctuations in GDP around “trend” are called business cycles • No one really knows for sure why this happens • Naturally, there are many competing theories Schumpterian View • The process of productivity growth is naturally a random process (a source of shocks) • Technological innovation is a process of “creative destruction” — innovations typically favor some sectors over others — even in favored sectors, innovations are slow to diffuse (so productivity gains are in the future) — GDP may decline in the short-run, as economy adjusts (e.g., retraining, reallocation, etc.) to exploit future productivity gains • Moreover, not all innovations turn out as planned — expectations over future productivity will evolve randomly as information is updated (information shocks) — actual productivity may be higher or lower than expected (a positive or negative productivity shock) — in any case, expectations are endogenous; they react “rationally” to exogenous information events Keynesian View • The process of productivity growth is naturally a relatively smooth process (not a major source of shocks) • Shocks stem from psychological sources that alter expectations: animal spirits (consumer confidence, business sentiment, etc.) — expectations are exogenous; they are proactive, rather than reactive • Expectation shocks may be either “rational” or “irrational” — rational: an exogenous change in expectations that changes the underlying fundamentals (self-fulfilling prophesies) — irrational: an exogenous change in expectations that bears no relation to (and so inconsistent with) underlying fundamentals • In any case, animal spirits are viewed as influencing economic activity through changes in “aggregate demand” Model 1: A Basic Schumpeterian Model • A representative agent with preferences ( ) • An aggregate production function = ; where denotes output, employment, and is an exogenous productivity parameter (equal to the real wage, in a competitive economy); Time constraint is + = 1 • Assume that fluctuates randomly over time (say, around some trend); let ∈ { } with • Solution satisfies (∗ 1 − ∗) = • Assume that SE dominates WE, so that ∗() is increasing in Implications • Output and employment varies over time; Employment and productivity is procylical (tends to move in the same direction as output); Broadly consistent with data • Allocation is efficient: no role for government policy • Reason: although welfare declines during a recession, economy reacts to exogenous decline in in the best manner possible • In fact, government stabilization policies are generally welfare-reducing Model 2: Animal Spirits • A representative agent with preferences ( ) • An aggregate production function given by = (n) where n is aggregate employment and is individual employment • Assume that (n) is increasing in n (an individual expects a higher return to their own labor , if everyone else is working harder); e.g. = ( if n ≥ if n where is a parameter (if aggregate employment is below individual productivity is low; and vice-versa) • Choice problem is now more complicated because individuals must try to forecast n • Conditional on some forecast n their choice problem is max ((n) 1 − ) • An individual’s optimal choice of now depends on their forecast of n; write this solution ∗(n) • Individual solution satisfies ((n)∗ 1 − ∗) = (n) • Of course, as everyone is identical in this economy, in equilibrium it must be the case that ∗(n) = n • And so we have one equation and one unknown determined by ((n)n 1 − n) = (n) • Interesting fact: there can be more than one solution to this equation — there may exist multiple equilibria — which equilibrium occurs is determined entirely by initial expectations of n (when people are formulating their choice problems) — these initial expectations are not determined by economic fundamentals; they are determined by psychology (animal spirits) • Imagine, for example, that individuals are (for some unexplained reason) “optimistic” — i.e., they expect n ; so they expect = — then solution is ( n 1 − n ) = • Imagine, instead, that individuals are (for some unexplained reason) “pessimistic” — i.e., they expect n ; so they expect = — then solution is (n 1 − n) = • Now imagine that expectations fluctuate randomly (waves of optimism and pessimism) Implications • Output and employment varies over time; Employment and productivity is procylical (tends to move in the same direction as output); Broadly consistent with data • Allocation is inefficient: potential role for government policy • Reason: although people are making individually-rational choices based on their beliefs, and while beliefs are rational (consistent with fundamentals), there is no good reason for beliefs to fluctuate in the first place • Government policy that stabilizes output at high level = n is desirable (if possible) Discussion • Two very different theories that deliver very similar (in fact, identical) predictions • Predictions are broadly consistent with observation; how can we discriminate between the two? • An important question because (unfortunately, or interestingly) the two theories deliver very different policy implications • One way is to try to evaluate consequences of historical policy interventions • But because reality is complicated, it is difficult to isolate effects in the data (many other things are changing at the same time) • In the meantime, people choose the theory that does least violence to their preconceived notions; e.g., Model 3: Uncertainty and Rational Expectations • Same as Model 1, except... • let ( | ) denote probability of conditional on information (news) ∈ { } and assume ( | ) ( | ) ( | ) ( | ) • i.e., good news means that is more likely and bad news means is more likely • assume that people know the probability structure ( | ) — i.e., they will form “rational” expectations based on all available information — note: expectations will react (rationally) to exogenous news events • Timing is as follows — news arrives (good or bad) — individuals make employment decisions based on this news (and their expectations of future productivity) — productivity is then realized (high or low) — output (and consumption) is determined by past employment decision and current productivity • Choice problem (conditional on news ) is max ( | ) ( 1 − ) + ( | ) ( 1 − ) • Solution is ∗() • If is linear in then solution is characterized by where ( ())∗ 1 − ∗) = () () = ( | ) + ( | ) which is expected productivity conditional on news • Note: () (); so ∗() ∗() Implications • Good news means high expected return to labor; employment boom (and vice-versa when news is bad) • Let () = ()∗() denote expected output; so good news means that expected output is high; i.e., () () • But actual output may turn out to be high or low, depending on realization of ∈ { } • Let ∗( ) = ∗() denote actual output; then ∗( ) () ∗( ) ∗( ) () ∗( ) • Output and employment fluctuate in manner consistent with observation • It appears that “mistakes” are made; e.g. — employment boom based on “optimism” that subsequently turns out to be wrong (productivity turns out to be low) — employment bust based on “pessimism” that subsequently turns out to be wrong (productivity turns out to be high) • These “mistakes” are consistent with rational expectations (must distinguish between ex ante and ex post) • Equilibrium is efficient; no role for government policy Model 4: Uncertainty and Irrational Expectations • This is probably the closest to what many policymakers have in mind • Same as Model 3 except... • Assume that actual productivity always remains constant; say 0 • But continue to assume that people form expectations based on ( | ) — interpret as “animal spirits” (i.e., good spirits, bad spirits) — these expectations are irrational; i.e., they are inconsistent with the underlying fundamentals (a constant 0) • People will react to “animal spirits” when they should not; employment will boom and bust as a consequence • But productivity always turns out to be 0; people never learn, even though they’re always wrong • Output and employment fluctuate in manner consistent with data • A definite role for government stabilization policy (assuming that policymakers are somehow more rational than their constituents) Application: Interpreting Output Gaps • Potential GDP: a theoretical concept describing the level of GDP that would transpire if markets worked “perfectly” • Output Gap: the difference between actual and potential GDP • Problem: potential GDP is not observable (assuming that it even exists) — so people try to estimate it, usually by drawing a “trend” line through the data • Problem: drawing a trend line through the data does not mean that it represents potential GDP — even worse, interpreting the measured output gap as evidence of market failure is highly dubious • Note: can identify potential GDP and output gaps in both Models 3 and 4 — markets work perfectly in Model 3, but not in Model 4