Annex M — Empirical protocol for the distributive tracing of global monetary easings

Annex 13 of 17 · ≈ 8 min read · 23 paragraphs · section 35 of 40 · search the book

Incorporating a review of the international literature · state of sources: July 2026.¶ 1

M.0 State of the international literature

This section is new and conditions all that follows. The distributive question of quantitative easing is one of the most heavily worked questions in the monetary macroeconomics of the last decade. A protocol that presented it as new would be set aside at once; a protocol that positions itself within a documented disagreement may, for its part, bring something.¶ 2

M.0.1 Three positions, and a disagreement over the sign

Egalitarian or neutral position. A substantial part of the work, mostly produced by central banks and international financial institutions, concludes that easing reduces income inequality or has no notable effect. The mechanism invoked is the labour-income heterogeneity channel: easing lowers unemployment, and the poorest households find employment again, which compresses the distribution. The work of the European Central Bank based on household wealth surveys finds this channel dominant and the effect on wealth inequality negligible. Heterogeneous-agent models and analyses using aggregate data often converge in the same direction, with weak and transitory net effects, dissipating within a few quarters.¶ 3

Inegalitarian position. Other work concludes the opposite. On matched Danish administrative data, the asset valuation gains induced by easing accrue far more to households endowed with wealth and greatly exceed the wage gains of households of modest means. A panel study of forty-nine countries between 1999 and 2019, mobilising local projections and augmented inverse probability weighting to treat endogeneity, concludes that there is a net relationship between asset purchase programmes and a rise in inequality.¶ 4

Non-linear position — the most instructive. Recent central bank work partially reconciles the two camps through a pattern of non-linearity: unconventional policies benefited all households, reduced inequality within the bottom ninety per cent through the fall in unemployment, and simultaneously widened the gap between the top decile and the rest through the rise in profits and in equity prices. The same literature indicates that forward guidance amplifies both aggregate and distributive effects.¶ 5

M.0.2 The alternative-policy counterfactual

The most disturbing result of the benchmarking exercise is the following: according to recent central bank work, had the policy rate not been constrained by its lower bound, a conventional rate policy would have produced more unfavourable distributive effects than quantitative easing. In other words, comparing easing with the absence of easing is not the right comparison.¶ 6

Rule. The protocol requires at least two counterfactuals of a different nature. The first is the absence of easing, as a polar reference. The second is the available alternative policy: an unconstrained rate path, or any instrument that the central bank could actually have mobilised. To publish the first without the second is misleading, because the reader will infer from it a judgement on the instrument whereas the calculation bears only on its existence. This requirement is a blocking one: without a second counterfactual, no distributive conclusion is published.¶ 7

M.0.3 Selection of the reference works

Note of caution. This table summarises dominant orientations, not established truths; each family contains dissenting work. The protocol requires that the review be redone, systematised and published with its inclusion criteria before any estimation — a narrative review does not suffice to found a pre-registration.¶ 8

M.0.4 The real gap, and the possible contribution

The benchmarking exercise reveals a striking asymmetry. The distributive literature on easing is almost entirely domestic: United States, euro area, United Kingdom, Denmark, with a few panels of advanced countries. The cross-border literature, for its part, bears on macroeconomic and financial effects — global financial cycle, bank risk-taking channel, capital flows — but very rarely on the question of who, in the recipient countries, captures the gains of an easing decided elsewhere and who bears its reversal.¶ 9

This reorientation has a consequence for the scope. The protocol does not seek to establish the domestic distributive effect of easing in the source jurisdictions: it borrows it from the literature, with its disagreements, and uses it as an input. Its own object is the chain that runs from the monetary operation to the final beneficiary situated outside the source jurisdiction, and the asymmetric treatment of gains and of reversals between the two.¶ 10

M.1 What “100 units injected” means

The common formulation of a monetary injection is useful for communicating and insufficient for attributing a benefit. Asset purchases increase the assets of the central bank and, as the counterpart, its liabilities, principally the reserves of depository institutions. Public data make it possible to observe the securities held, the operations and the reserves; they do not deliver the identity of the final economic beneficiary.¶ 11

Five normalisation bases are possible and do not measure the same thing: announced purchases, gross settled purchases, net settled purchases, reserves created, and identified monetary surprise. The protocol retains net settled purchases as the principal base, gross purchases as a first-level account, reserves as a coherence check, the announcement for announcement analysis alone, and the identified surprise as a complementary econometric base. Every “per 100” presentation declares its base; mixing two bases in the same table is an error of account, not an approximation.¶ 12

M.2 Non-additive registers and measurement by rank

Four registers coexist and are never added together: the operation and its settlement; asset valuation and financial income; the cost of financing, credit and balance sheets; macroeconomic effects, including employment and labour income. The fourth register is the one that the central bank literature finds dominant for income inequality; its omission would produce a regressive result by construction and not by measurement.¶ 13

Measurement by rank, not by index. Results are reported by quantile and by functional group — asset holders, borrowers, wage earners, pensioners — and never consolidated into a single index. The reporting structure must make it possible to distinguish an equalising effect within the bottom ninety per cent from a widening effect vis-à-vis the top decile, since that is the pattern which the recent literature identifies and which an aggregate index would erase.¶ 14

Income and wealth kept separate. The two dimensions behave differently depending on the study, and their conflation explains an important part of the contradictory conclusions of the field. No result combines the two in a single magnitude.¶ 15

M.3 Identification

The shocks must separate policy proper, the information revealed by the central bank about the state of the economy, and the reaction to the crisis. This separation is identical to that required by the shock vector protocol, and the two annexes share the same register of shocks in order to avoid two divergent measurements of the same event.¶ 16

The protocol retains a panel design with local projections, predetermined exposures, and explicit treatment of the endogeneity of the decision to ease — central banks ease when the economy deteriorates, so that a naive comparison would confound the effect of the policy with that of the recession which motivated it. The recent international panel literature proposes weighting methods suited to this problem; they are taken up here as the reference specification rather than reinvented.¶ 17

M.4 The full cycle and asymmetry

The most pertinent question for the present work is not the effect of easing taken in isolation, but the symmetry of the cycle: are the agents who benefit from the expansion those who bear the tightening? The literature on the unwinding of asset purchases indicates a state dependence and a heterogeneity between households which forbid presuming symmetry. This attention to the distribution between agents extends the analysis of the redistribution channel of monetary policy (Auclert, 2019).¶ 18

Rule of analysis. The gains of expansion and the costs of reversal are reported separately, by initial beneficiary and by date, and are never set off against one another without a documented transfer. A presumed intertemporal offsetting would amount to assuming precisely what the analysis must establish. This rule is identical to the one that governs the treatment of counterparties in the measurement layer.¶ 19

M.5 Stopping rules

M.6 Antithesis and rejection criteria

The strongest antithesis is that this annex adds a layer of complexity to a question already densely treated, without new data or superior identification, and that its most likely conclusion is to recover the existing results with wider uncertainty bands. The objection is well founded for the domestic scope, and the protocol accepts it: it claims no contribution of its own there.¶ 20

It is less so for the cross-border scope, where the question has not been posed in distributive terms. The protocol must nevertheless be rejected, or strongly reduced, in four cases: if the data of the recipient countries permit no breakdown by rank, in which case it is limited to the register of the operation and of the cost of financing; if the cross-border results prove entirely derivable from the domestic results already published; if the asymmetry of the cycle is not detectable with the available horizons; or if the protocol produces no result capable of contradicting the initial intuition of the present work, which would signal an arrangement built to confirm.¶ 21

M.7 Graduated compliance

M.8 What this annex brings to the present work

It brings three things, and one must be precise about what it does not bring. It brings first a discipline of language: it forbids the formulation of a distributed injection, which is technically false and which exposes the document to immediate refutation by any monetary economist. It brings next the symmetry of the cycle, a question to which the measurement layer is already devoted for the tightening and which remains incomplete so long as easing is not treated according to the same rules. It brings finally a displacement of the distributive question towards the recipient countries, where it has not been posed.¶ 22

It brings no conclusion on the domestic effect of easing in the source jurisdictions, and must not be presented as such. On this ground, the present work borrows from a divided literature, and the most solid way of citing it is to cite its division rather than the camp that suits the argument.¶ 23

@incollection{yaiche_hc_annexe_m,
  author    = {Yaïche, Nizar},
  title     = {Annex M — Empirical protocol for the distributive tracing of global monetary easings},
  booktitle = {The Humanity Compact},
  year      = {2026},
  version   = {2.3},
  doi       = {10.5281/zenodo.22746533},
  publisher = {Zenodo},
  url       = {https://humanitycompact.org/ouvrage/annexe-m/}
}

A book by Nizar Yaïche, former Minister of Finance of the Republic of Tunisia. Who is the author?