Contribution
It carries the economists’ increasing-returns account of path dependence (Arthur’s four conditions and North’s extension of them to institutions) into political science, and argues that it applies there more, not less. It names which features of politics make increasing returns more prevalent (three) and which make them more intense (three), so that “path dependence” becomes a claim about mechanisms rather than a synonym for “history matters”. It reports no data; its evidence is the economics literature and one sketched case.
Key insight
The objection that kills most path-dependence stories in economics, that markets and far-sighted investors correct inferior paths, does not carry over to politics. Political actors cannot exit compulsory rules, act collectively, cannot measure performance well, discount the future at the electoral cycle, and work inside institutions deliberately built to resist change. So “path dependence arguments are likely to be even more relevant to an understanding of politics than they are in other areas of the social sciences” (p. 10).
Assumptions
- Increasing returns as the source. Path dependence is taken in the sense of a Pólya urn: chance early events plus a rule that ties later probabilities to earlier outcomes. Arthur’s four properties (unpredictability, inflexibility, non-ergodicity, potential path-inefficiency) and his four conditions (large set-up costs, learning, coordination effects, adaptive expectations) are taken over without re-derivation (pp. 11–16).
- Institutions include public policies. “Although unorthodox, the inclusion of public policies as well as formal institutions in this formulation is important” (p. 24).
- Stylised contrast with markets. The market side is the textbook one (exit, prices, repeated purchase, capital markets); footnotes 17 and 18 grant the simplification.
- Mechanisms from evolution are set aside. Footnote 7 prefers economics to Gould’s punctuated equilibria because “Most aspects of politics lack anything like the mechanism of natural selection”, with international relations and some features of electoral systems as exceptions.
Key results
There are no new data. By part:
- I. Increasing returns processes. The Pólya urn and Arthur’s four properties, plus sequencing: “Early events, including “noise”, matter much more than later ones” (p. 12).
- II. Economics. Technology lock-in (QWERTY, light-water reactors, VHS), economic geography (Krugman), trade, endogenous growth and North on institutions. Against Liebowitz and Margolis, the remediability test is called “a debater’s point” (p. 22) when the aim is to explain rather than to advise, and their reliance on far-sighted markets is said to have “very limited relevance for political scientists” (p. 23).
- III. Why politics is more prone. Three features make increasing returns more prevalent: institutional density (“Much of politics, in other words, is based on authority rather than exchange”, p. 25), collective action, and complexity and opacity. “Increasing Returns in fact characterize three of the most prominent features of political environments: processes of institutional development, processes of collective action, and processes of social interpretation” (p. 30). Three features make them more intense: “the weakness or absence of efficiency-enhancing mechanisms of competition and learning; the shorter time horizons of political actors, and the greater “stickiness” of political institutions” (p. 24).
- Social interpretation. After North and Denzau, “Mental maps are subject to increasing returns” (p. 29), and basic outlooks, “ranging from ideologies to understandings of particular aspects of governments or orientations towards political groups or parties”, are tenacious and so path dependent (p. 29).
- Time horizons. “Because the decisions of voters, which determine political success, are taken in the short-run, elected officials employ a high discount rate” (p. 32). Leaving an inferior path costs now and pays later: “The costs of doing so are borne in the short-run, while the benefits will generally only accrue in the long run” (p. 33). Pierson calls the neglect of this a gap: “Yet we know relatively little about the time horizons of different political actors” (p. 32). Accountability is weakened “because of the relatively rapid turnover of key positions” (p. 32). Footnote 21 grants that an actor with a longer horizon, such as an interest group, can make the long term count to politicians, and doubts that it works as well as capital markets.
- Stickiness. Institutions are made hard to change on purpose: to bind successors (Moe’s political uncertainty) and to bind oneself (credible commitment). Unanimity in the European Union and supermajorities for the US constitution are the examples (p. 34).
- The case. The Clinton health plan (1993–94) and the Republican Medicare and Medicaid cuts (1995–96) both failed against the interests that had grown up around existing policy and against veto points (pp. 35–36). It is an illustration, chosen after the fact.
- IV. What path dependence implies. Multiple equilibria, contingency, timing and sequence, and punctuated equilibria: “rather than exhibiting gradualism, the phenomenon under investigation may be subject to long periods of relative stability followed by brief bursts of rapid change” (p. 36). “In path dependent processes, the necessary and sufficient conditions for current outcomes may have occurred in the past” (p. 39). Uses: a corrective to functionalism, sharper accounts of critical junctures, hypotheses about the conditions of change, and humility about prediction: “The problem lies not in our methods, but in the character of the political world itself” (p. 41).
- Limits granted. “Not all aspects of political life are subject to increasing returns” (p. 41). Often the concept will mainly serve “a “negative” purpose”, a caution against reading outcomes as inevitable or functional (p. 10).
Claims
The claims of the 1997 working paper. The 2000 article was not read.
| id | claim | strength | support |
|---|---|---|---|
| C1 | Arthur’s conditions for increasing returns (set-up costs, learning, coordination, adaptive expectations) hold for institutions and for many social interactions, not only technologies | moderate as a synthesis | Arthur 1994, North 1990; Part II |
| C2 | Increasing returns are more prevalent in politics than in markets, because of institutional density, collective action and opacity | weak: argued, untested | Part III; stylised contrast with markets |
| C3 | Lock-in is more intense in politics, because competition and learning correct weakly, horizons are short and institutions are deliberately sticky | weak: argued, untested | Part III; Moe, North and Weingast, the credible-commitment literature |
| C4 | Elected officials discount the future at the electoral cycle, so they will choose paths with early payoffs and will rarely leave a path whose exit costs come now | weak to moderate: a standard political-economy argument with one anecdote (Stockman) | p. 32–33; Figure 1 reasoning |
| C5 | Shared interpretations of politics (mental maps, orientations to parties) are themselves subject to increasing returns | weak: asserted from North and Denzau | p. 29 |
| C6 | Path-dependent change is punctuated: long stability, brief bursts, often at exogenous shocks and tipping points | weak: stated as an implication | Part IV |
| C7 | Path dependence corrects functionalist explanation: an institution’s persistence does not show that it serves a function | moderate as an argument | Part IV |
Concepts
- Increasing returns / positive feedback: each step down a path makes that path more attractive for the next step.
- Path dependence: here specifically the outcome of an increasing- returns process, not the weak sense that later events depend on earlier ones (which Liebowitz and Margolis call first-degree).
- Remediable path dependence: Liebowitz and Margolis’s third degree, where a feasible better path existed; Pierson thinks the test asks too much for explanation.
- Institutional density: the share of a domain governed by compulsory rules, formal institutions and public policies.
- Stickiness: deliberate resistance of political arrangements to change, built to bind successors or oneself.
- Time horizon / discount rate: how much an actor weighs consequences beyond the next payoff; for elected officials, the electoral cycle.
- Historical causes (after Stinchcombe): an ordering moment that set a pattern which is now reproduced although the moment no longer recurs.
Connections
It builds on Arthur (1994) for the mechanism, on North (1990) for the step to institutions, and on Moe (1984, 1990) for what is distinctive about politics. It sets itself against Liebowitz and Margolis (1995) and places itself as the micro-foundation for historical institutionalism: Collier and Collier (1991), Skowronek (1993), Lipset and Rokkan (1967), Ikenberry (1994). The record holds neither book Pierson cites, but it holds Arthur’s 1989 article as LIT-1314 (THEORY-863, Active) and North’s 1991 article on institutions as LIT-489 (THEORY-429). Acharya, Blackwell and Sen (LIT-1388, NOTE-1089) cite the 2000 article for “the path dependence of historical institutions over time”.
Bearing on the record
- CLAIM-423 (timescale and authority in a political collective). The
claim’s own defeater asks whether Pierson already states it. On the 1997
text he does not. Two pieces are close:
- Time horizons as a variable. Pierson makes the horizon of a political actor a cause of which paths are taken and kept: elected officials employ “a high discount rate”, interest groups may have longer horizons (fn. 21), and rapid turnover weakens accountability (p. 32). That is CLAIM-423’s raw material: components with different horizons behave differently. But he uses horizons to explain lock-in, not to explain which component holds authority over which. He never sets an administration’s memory against an elected government’s power to change the law, which is CLAIM-423’s point.
- Authority as the stuff of politics. “Much of politics, in other words, is based on authority rather than exchange” (p. 25), and institutions are built so that designers “purposely create structures that even they cannot control” (after Moe, p. 34). Stickiness is how a past coalition holds authority over later ones. That supports CLAIM-423’s clause that a slower component (law, treaty) can govern a faster one, and says why: deliberate design, not slowness.
- So the claim survives as the record’s own, and Pierson is its nearest source for the horizon half. He explicitly calls the time horizons of different political actors an open question (p. 32). Whether the 2000 article adds anything on the use of authority itself was not checked.
- CLAIM-331 (higher levels are slower). No direct bearing. Pierson’s slow things (institutions, policies, outlooks) are slow because of increasing returns and design, not because they are higher levels.
- CLAIM-184 (anchors entrench), CLAIM-277 (trust entrenches what gained influence early), THEORY-429 (North). Pierson is the political-science statement of the same mechanism. Nothing here tests them.
- CASE-133 (party realignment). Two points, one each way. For: C5 makes “orientations towards political groups or parties” path dependent, the kind of durable outlook the case is about. Against a casual reading: Pierson’s path dependence implies punctuated change, while the case, following Schickler, is gradual change from below over decades. The case does not cite Pierson, so nothing is misattributed; but path dependence is not the frame the case’s own account uses.
- CLAIM-327 (a slow collective variable read and regenerated). C5 is the political instance: shared interpretations feed back on what is noticed, “Confirming information tends to be incorporated, while disconfirming information is filtered out” (p. 29).
- THEORY warranted? One, Proposed: THEORY-885, C2 with C3 and C4, which the record does not hold. THEORY-429 holds North’s claim for institutions in general; Pierson’s claim is that politics as a domain is more prone than markets, and why.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- An argument with no data and one illustration chosen after the event. Nothing in it shows that a political outcome was locked in by an early contingent event rather than by persisting causes.
- The comparison with markets is stylised. Markets have institutions, collective action and opacity too, and the paper grants the qualifications in footnotes rather than measuring the difference.
- It does not say how to tell a path-dependent process from a persisting cause in a given case. “history matters” in Liebowitz and Margolis’s first-degree sense is what most of its illustrations show.
- The 1997 version is the one read. The APSR article is a revision and may differ in structure and in its list of features.
Open questions
- How could the prevalence claim (C2) be tested? A comparison of reversal rates for policies and for comparable market arrangements, with set-up costs and coordination held comparable, would bear on it.
- Do longer-horizon actors (courts, bureaucracies, interest groups) systematically hold authority over shorter-horizon ones, and under which institutions? Pierson names the question and leaves it; it is CLAIM-423’s.
- When does an increasing-returns process end? The paper explains lock-in better than the bursts of change it predicts.