Editorial view of a dark hydraulic macroeconomic control room

Epoch 02 · 1936–1980 · Policy Game

The Economy Learns the Rule

Economists built a policy machine—then discovered that the machine reacts to its operator.

Your roleDecision-maker

Change a transparent macro rule and inspect the local phase field.

The driving question

Why can an economy settle below full employment—and why do policy levers stop being stable?

Economists made the economy as a whole visible and steerable—then discovered that uncertainty, dispersed information, expectations, and policy feedback kept changing the machine.

Illustrative macro modelNot a historical estimate

Output gap0.0

Inflation2.0

Unemployment5.0

The policy board

You are the decision-maker

Steer the town. Then change the rule.

Move one lever at a time. The model exposes its assumptions instead of hiding them behind a dashboard.

Demand support0.0Interest-rate stance0.0Inflation shock0.0Expectation sensitivity35%

y = demand(1 − 0.55e) − 0.8rate − shockπ = 2 + shock + 0.55y + 0.35e(shock + demand)u = 5 − 0.45y

Dimensionless teaching parameters. This model is neither an estimate nor a forecast.

Three rules the operator cannot hold fixed

The machine works. Then its parameters answer back.

Solow changes the destination. Friedman changes the short-run menu. Lucas changes the coefficient itself. Each instrument keeps the policy operator—and the boundary of the model—in view.

Instrument 06 · Robert M. Solow · 1956

Capital deepening raises the level. Then the curve runs out of slope.

Investment adds capital, but depreciation subtracts from a stock whose extra units produce progressively less additional output.

Move the saving rate and the destination changes. The economy approaches a higher capital level, not permanent acceleration. Solow disciplined accumulation precisely by leaving sustained technological progress outside the machine.

Production: y = k^0.35
Depreciation: 8% per period
Technology: Held fixed

Open Solow’s 1956 paper ↗

Figure 02 · Transition line

More saving raises the destination, not the long-run slope.

At a 18% saving rate, capital approaches an illustrative steady state of 3.48 per worker.

Saving rate18% · steady-state k 3.48

Period 60 · saving 18%k 3.33 · y 1.52

Exact representative values and equation

Period	Capital	Output	Investment	Depreciation
0	0.60	0.84	0.15	0.05
10	1.57	1.17	0.21	0.13
20	2.30	1.34	0.24	0.18
30	2.76	1.43	0.26	0.22
40	3.05	1.48	0.27	0.24
50	3.23	1.51	0.27	0.26
60	3.33	1.52	0.27	0.27

y_t = k_t^0.35; k_(t+1) = k_t + s y_t - 0.08 k_t; k* = (s / 0.08)^(1 / (1 - 0.35)).

Illustrative capital and output per worker; model period is not calendar time. The Cobb-Douglas parameters, saving rate, depreciation rate, and periods are authored teaching values. Technology and population are held fixed; the path is not a country estimate or forecast.

Instruments 07–08 · Phillips 1958 / Friedman 1968

The menu moves while policymakers order.

A surprise can move unemployment along one short-run curve. Repetition changes expected inflation—and therefore changes the curve itself.

Phillips documented British money wages. Later policy debate converted that historical pattern into an inflation menu. Friedman’s objection was not that the short run vanishes; it was that the menu cannot be ordered indefinitely without teaching people the price.

Open Phillips’s paper ↗Open Friedman’s address ↗

Figure 03 · Expectations-augmented Phillips curves

The first move is real in the model. It is not permanent.

At period 0, expected inflation is 2.0% and illustrative unemployment is 3.0%.

Actual inflation6.0% · initial surprise 4.0 pp

Period 0 · expected inflation 2.0%Unemployment 3.0%

Exact period values and equation

Period	Actual inflation	Expected inflation	Surprise	Unemployment
0	6.0%	2.0%	4.0 pp	3.0%
1	6.0%	4.0%	2.0 pp	4.0%
2	6.0%	5.0%	1.0 pp	4.5%
3	6.0%	5.5%	0.5 pp	4.8%
4	6.0%	5.8%	0.3 pp	4.9%
5	6.0%	5.9%	0.1 pp	4.9%
6	6.0%	5.9%	0.1 pp	5.0%

u_t = 5 - 0.5(π - π^e_t); π^e_(t+1) = π^e_t + 0.5(π - π^e_t).

Illustrative unemployment and inflation rates in percent; model period is not calendar time. Phillips studied UK money-wage changes and unemployment; this teaching model uses inflation only to show Friedman's expectations argument. It does not reproduce Phillips's observations or estimate a modern Phillips curve.

Instrument 10 · Robert E. Lucas Jr. · 1976

The controller is inside the system.

A coefficient estimated when policy surprises people is not a mechanical constant waiting to be reused under a public rule.

Move the share of the policy rule that agents anticipate. The impulse stays fixed; the behavioral response changes. The numerical slope is synthetic. The conceptual warning is Lucas’s: policy evaluation needs relationships that survive the intervention.

Open Lucas’s 1976 critique ↗

Figure 04 · Regime coefficient slopegraph

The lever is unchanged. Its coefficient is not.

With 50% anticipation, the illustrative response coefficient is 0.47 and the fixed impulse produces 1.42 index points.

Anticipated share of policy rule50% · β 0.47

Impulse fixed at 3 · anticipation 50%Response 1.42 index points

Exact regime values and equation

Regime	Anticipation	Coefficient	Response to impulse 3
Surprise	0%	0.80	2.40
Current	50%	0.47	1.42
Anticipated rule	100%	0.15	0.45

β(a) = 0.15 + (0.80 - 0.15)(1 - a); response = β(a) × 3, where a is the anticipated share of the policy rule.

Illustrative output-response index per one-point policy impulse. The coefficient endpoints, interpolation, and policy impulse are synthetic. The figure demonstrates regime dependence; it neither estimates a structural parameter nor claims that anticipation is the only channel through which policy behavior changes.

Failure → repair

Expectations and regime change make the coefficients respond to the operator; the Lucas critique closes the room.

The equations are illustrative and dimensionless, never presented as an estimate or forecast.

The canon · Ten landmark works

Read the contribution. Then read the boundary.

Chronological, not ranked. Influence records intellectual reach—not endorsement or empirical validation.

011937

John R. Hicks

Mr. Keynes and the ‘Classics’; A Suggested Interpretation

Journal article

What it made visible

IS–LM goods–money equilibrium

Why it mattered

Hicks compressed part of Keynes into simultaneous equilibrium in the goods and money markets. IS–LM became the teachable control panel of the postwar Keynesian synthesis.

The limit

Its tractability removes much of Keynes’s time, finance, and radical uncertainty; the usable panel is narrower than the economy it depicts.

Read the primary record ↗

021939

Jan Tinbergen

Statistical Testing of Business-Cycle Theories. II: Business Cycles in the United States of America, 1919–1932

League of Nations Economic Intelligence Service monograph, 244 pages

What it made visible

A dynamic macroeconometric system of linked behavioral equations estimated from time-series data

Why it mattered

Tinbergen turned competing business-cycle theories into an estimated model of the whole US economy. This was the decisive step from drawing macro mechanisms to constructing an empirical machine that could reproduce fluctuations and support stabilization experiments.

The limit

The equations left simultaneity, identification, lag choice, and coefficient stability unresolved; Haavelmo supplied the probabilistic foundation and Lucas later exposed policy-regime dependence.

Read the primary record ↗

031944

Trygve Haavelmo

The Probability Approach in Econometrics

Monograph supplement

What it made visible

Probability, identification, and simultaneous equations

Why it mattered

Haavelmo gave econometrics an explicit probability foundation and separated structural relations from equilibrium correlations. His framework made estimation, testing, identification, and causal policy analysis part of one system.

The limit

The machine gains sensors and error bars, but intervention still assumes structural relations remain autonomous when policy changes.

Read the primary record ↗

041945

Friedrich A. Hayek

The Use of Knowledge in Society

Journal article

What it made visible

Dispersed knowledge and prices as signals

Why it mattered

Hayek recast allocation as a problem of local, tacit, and dispersed knowledge rather than optimization with all data given. Prices coordinate separate plans without first assembling every fact in one place.

The limit

No central dashboard observes the complete state of the economy; this information-processing problem prepares the digital turn.

Read the primary record ↗

051954

Kenneth J. Arrow and Gérard Debreu

Existence of an Equilibrium for a Competitive Economy

Journal article

What it made visible

General competitive equilibrium

Why it mattered

Arrow and Debreu proved equilibrium existence for an integrated system of production, exchange, and consumption under explicit assumptions. Their result became the rigorous benchmark for welfare, finance, and later theories of market failure.

The limit

Existence is not uniqueness, stability, computability, or realism; later economics works by relaxing the theorem’s demanding gears.

Read the primary record ↗

061956

Robert M. Solow

A Contribution to the Theory of Economic Growth

Journal article

What it made visible

Diminishing returns and balanced growth

Why it mattered

Solow replaced knife-edge fixed proportions with factor substitution and diminishing returns, making stable balanced growth possible. Capital deepening raises output levels, while sustained per-capita growth ultimately requires technological progress.

The limit

The growth engine stabilizes, but its decisive long-run fuel—technology—remains outside the model, setting Epoch 3’s central question.

Read the primary record ↗

071958

A. W. Phillips

The Relation Between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861–1957

Empirical article

What it made visible

The empirical wage Phillips curve

Why it mattered

Phillips documented a nonlinear inverse relation between unemployment and money-wage changes in British data. Later economists transformed his tentative finding into an apparent inflation–unemployment policy menu.

The limit

A historical correlation becomes a tempting control surface, but supply shocks and expectations can move the surface while policy uses it.

Read the primary record ↗

081968

Milton Friedman

The Role of Monetary Policy

Presidential address

What it made visible

Natural rate and expected inflation

Why it mattered

Friedman argued that expected inflation removes any permanent unemployment gain from monetary expansion, leaving only a short-run trade-off. Alongside Edmund Phelps, he made expectations and variable policy lags central to stabilization.

The limit

The target learns: repeated use of the same lever changes expectations, weakens its real effect, and can turn stabilization into disturbance.

Read the primary record ↗

091970

George A. Akerlof

The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism

Journal article

What it made visible

Asymmetric information and adverse selection

Why it mattered

Akerlof showed that private quality information can drive good products out and shrink or destroy beneficial trade. Warranties, brands, screening, and reputation became endogenous responses to information failure rather than peripheral details.

The limit

Agents do not share one state of the world; information becomes an economic input and institutions become information-processing devices.

Read the primary record ↗

101976

Robert E. Lucas Jr.

Econometric Policy Evaluation: A Critique

Journal article

What it made visible

Policy-regime dependence and structural invariance

Why it mattered

Lucas showed that parameters estimated under one policy regime may change under another because decisions incorporate expectations about the rule. Credible counterfactuals therefore require structural behavior rather than extrapolated reduced-form correlations.

The limit

The controller is inside the system: a new policy can rewire the machine instead of merely moving it to another point.

Read the primary record ↗

Selection & evidence

A canon, not a leaderboard.

These ten works are ordered by first publication. Selection weighs paradigm effect, conceptual durability, downstream reach, cross-generational influence, and non-redundancy. Every entry names a contribution and a boundary.

Read the full literature paper →