--- lang: en title: >- Overview: The Systematic Reconstruction of Pension Allocation Under Sustained Low Interest Rates and the Decumulation Phase—A Diagnostic Framework Built from Five Progressive Papers description: >- This paper is the systematic overview of five papers. It rejects the false dichotomy of "reaching for yield vs. asset-liability management (ALM)"—a di --- ## 1. The Master Question: Why a Single Configuration-Layer Inquiry Requires Five Papers, Not One Over the past four decades, pension systems in developed economies have grown up under an implicit premise: that there exists a stable, exogenously given risk-free rate curve, against which liabilities can be discounted, assets can be priced, and duration can be matched. The entire actuarial science, the entire asset-liability management (ALM) paradigm, the entire "long money for long investment" asset-management creed, have all treated this curve as the foundation lying outside the institution itself. Once interest rates entered a prolonged low-rate regime, and the first generation of defined benefit (DB) plan members moved en masse into the payout phase (referred to throughout this paper as the "decumulation phase"), this premise was for the first time systematically shaken. The foundation itself began to shift, and the entire configuration logic built upon it began, in consequence, to expose assumptions that had never before been interrogated. Facing this situation, the mainstream responses in academia and industry fall broadly into two camps. The first is the "reaching for yield" narrative: the decline in interest rates depressed returns on safe assets, and pension funds, to make up the shortfall in returns, were forced to seek risk premia from equities, alternatives, and overseas assets—the aggressive turn in asset allocation is read as a first-order response to low rates. The second is the "asset-liability management" (ALM) narrative: the decline in interest rates, amplified through discounting, inflated the present value of liabilities, and pension funds, to hedge the duration gap, were forced to increase holdings of long-dated bonds, de-risk, and pursue buyouts—the conservative turn in asset allocation is read as a first-order response to liability pressure. These two narratives appear to be opposites—one says pension funds are adding risk, the other says pension funds are de-risking—yet they share the same unexamined master premise: **"the interest rate" is an exogenous scalar that enters the allocation function linearly through some channel (whether a preference channel or a liability channel), and allocation behavior is the dependent variable of this scalar.** The starting point of this research is precisely a rejection of this master premise. Once one acknowledges that the very manner in which "the interest rate" enters allocation is itself refracted and mediated, layer upon layer, by institutions, by accounting, and by governance and settlement-physics conditions, the dichotomy of "reaching for yield vs. ALM" is no longer an empirical question to be adjudicated by data, but a pseudo-dichotomy sharing a false premise. And the volume of work required to dismantle this pseudo-dichotomy simply cannot be compressed into a single paper. There are three reasons for this. First, the refraction occurs at more than one layer, and each layer has its own independent mechanism, its own independent identification problem, and its own independent political-economic origin. The market interest rate is first refracted by **accounting conventions** (the subject of Paper One): what enters the allocation function is never the market rate itself, but four "shadow rates" each refracted through four misaligned regimes of institutional accounting. Yet the divergence in accounting conventions is itself rooted in a deeper divergence in **institutional form** (the subject of Paper Two): DB and DC are not two discrete institutional types but two degenerate corner points on a continuous spectrum parameterized by "degree of risk-sharing," and so-called "de-risking," in welfare terms, is often not risk reduction at all but a one-off wealth transfer among coexisting generations along this spectrum. And whether an institution is capable of holding a given class of assets, of deepening its allocation, in turn depends on **governance control rights**—a capacity far harder to transfer (the subject of Paper Three): what can be outsourced is investment capability, but what cannot be migrated is the allocation of control rights over that capability; the reason "copying Canada's homework" cannot be replicated is precisely that one can learn the capability but not the control rights. These three layers are nested, and cannot be made clear within the space of a single paper—covering the refraction of accounting conventions, the institutional spectrum, and the endogeneity of governance all at once. Second, the effects of low interest rates are not confined to the accumulation and allocation end; they must also be re-examined within the **decumulation (payout) phase**—a newly emergent window of pressure (the subject of Paper Four). The 2022 UK liability-driven investment (LDI) episode is not a footnote illustrating "how interest rates hurt pension funds," but rather an event that revealed an entire set of previously concealed mechanisms: improvement in the funding line does not constitute a safety signal for the liquidity line; the localization of risk is, to a large extent, an attribution choice endogenous to each country's institutions; the thickening of buffers merely shifts the trigger point rather than eliminating the risk; and the widely circulated inference that "the Netherlands and Denmark had no crisis, so their institutions must be superior" is in fact a textbook case of survivorship bias. The decumulation phase re-weighs, within an environment of potentially lethal stress, the mechanisms identified in the first three papers at the accumulation end—and this re-weighing itself constitutes an independent corrective undertaking. Third, and most fundamentally, if the diagnoses of the first four papers are to be brought to bear on a late-developing economy that is simultaneously experiencing declining interest rates, accelerating population aging, and the multi-layering of its institutions, one must answer the technical judgment of "can this simply be copied" (the subject of Paper Five). The divergences that China's multi-pillar pension system exhibits across its five layers are, in essence, a layered misalignment of "who holds the property rights to which category of risk, and who has the capacity—not merely the obligation—to price and provision for it"; these five layers can be ranked and compared along several commensurable dimensions, but their allocation logics cannot be applied across layers, still less can overseas templates—premised on a stable risk-free anchor, mature annuitized markets, and transferable governance independence—be transplanted directly. The concluding paper both closes out each of the first four papers in turn and delivers the final accounting of the entire matrix's real-world implications. For this reason, this research is not a loose assemblage of five separate topics, but a single chain of diagnosis that proceeds from "what is the interest rate that actually enters allocation," descending layer by layer into institutions, governance, the decumulation phase, and finally into specific national cases. What follows first sketches the skeleton of this progressive logic, then identifies the core judgment running through the entire chain, and finally sets out the methodological self-awareness and boundaries of the study. --- ## 2. Progressive Logic: How the Five Papers Unfold Layer upon Layer This section explains, paper by paper, what each paper inherits from its predecessor, what it advances on its own, and what groundwork it lays for the paper that follows. The relationship among the five papers is not that of parallel chapters, but of five vertebrae deepening in sequence along the same diagnostic spine. ### Paper One, "The Interest-Rate Transmission Mechanism of Pension Asset Allocation under Prolonged Low Rates": What Enters the Allocation Function Is Not the Interest Rate, but a Shadow Rate Refracted through Institutional Accounting Paper One directly addresses the parent proposition, suspending the premise of the pseudo-dichotomy of "reaching for yield vs. ALM," and instead asking a more fundamental question: when we say "interest rates affect allocation," what exactly is it that enters the allocation function? Its core advance is to deconstruct the independent variable of the allocation function itself. After market rates decline, there is no direct causal chain running straight from interest rates to allocation; what exists instead is a translation mechanism — the market rate is refracted, through four mutually misaligned systems of institutional accounting (Dutch-style FTK/UFR discounting, IAS19R recognition under International Financial Reporting Standards, the U.S. Section 430 funding basis, and the deductibility basis under tax law), into a "shadow rate" specific to each; the wedge between these shadow rates opens as rates fall and closes as the decumulation (payout) phase draws near, and allocation drift is driven by the opening and closing of these accounting wedges rather than by the market rate itself. The so-called "interest-rate elasticity" that is observed is, to a considerable extent, a disguise for wedge elasticity. The existence and shape of this translation mechanism can be cleanly identified via a quasi-natural experiment: the 2013 elimination of the corridor method under IAS19R, which mandated recognizing pension gains and losses using a risk-free curve (rather than the sponsor's self-selected expected-return rate), constituted an exogenous "quasi-rate shock" for sponsors on international accounting standards, whose magnitude of de-equitization rose with the aggressiveness of the sponsor's prior accounting assumptions and with balance-sheet fragility, while being only weakly related to liability duration — this demonstrates that it is the accounting basis, rather than duration itself, that drives allocation. At the level of identification, Paper One simultaneously issues a serious warning: even setting aside the refraction of accounting bases, the naive "interest rate → allocation" regression coefficient is itself not to be trusted. On one hand it is contaminated by population aging as a common cause (aging projects onto both interest rates and allocation simultaneously via two channels — "life-cycle saving depressing the natural rate" and "plan maturity raising duration demand" — thereby manufacturing spurious correlation); on the other hand it is reversed by demand-side reflexivity (price-insensitive, rigid pension buying pressure self-suppresses the term premium and endogenously prolongs the low-rate environment, so that the arrow flips from "interest rate → allocation" to "allocation demand → interest-rate environment"). It is worth emphasizing that these two forces are explicitly located at different levels within Paper One's framework rather than competing with one another: the accounting wedge concerns the mechanism-level question of "how the market rate is translated into allocation via accounting bases" — its existence and shape — while the demographic common-cause concerns the identification-level warning of "why the observed interest rate–allocation coefficient cannot be read naively"; the two operate at different levels, do not contest the same residual variance, and therefore require no sign-reversing cross-section to separate them. This distinction of levels is itself a methodological contribution of Paper One: it refuses to conflate mechanism-level explanation with identification-level warning, thereby avoiding the inflation of apparent identification through an infeasible separating design. Paper One's honesty is likewise evident in its treatment of overseas allocation and the settlement-physics layer. Regarding overseas allocation, it corrects a baseline that had been treated as an established causal relationship — the supposedly monotonic relationship whereby a one-percentage-point decline in the domestic rate produces a corresponding rise in overseas asset allocation is in fact spuriously monotonic; the true constraint on overseas allocation is not the domestic interest-rate differential but the hedged carry, whose sign can run opposite to the domestic long-end rate, while the allocation ceiling is a corner solution manufactured by regulation (capital and matching penalties on hedged foreign liabilities endogenously determine the overseas-allocation ceiling), rather than the outcome of interior-point optimization within a return budget. Regarding the settlement-physics layer, it moves the hard constraint on allocation forward to the physical layer of "whether settlement can occur, whether leverage can be added": the same government bond is simultaneously an asset and collateral, and procyclical margin models mechanically raise margin and haircuts as volatility rises, forcing sales that further depress the bond price, forming a self-reflexive collateral loop; meanwhile cash variation margin re-prices duration hedging as an implicit short position in cash, constituting an "instrument-mismatch tax." Paper One honestly distinguishes which claims can truly be borne out by exogenous shocks (the translation mechanism of accounting bases, the collateral self-reflexivity and intermediation tax at the settlement end — the latter identifiable via a quasi-natural experiment using the expiry of clearing exemptions) from those that can be retained only as mechanism hypotheses or case studies (such as the distributive politics of "refusing to lower the discount rate" in public DB plans, which — because it cannot be separated purely econometrically from the ALM constraint under the accounting identity of the expected-return method — is explicitly downgraded to a qualitative case; or duration rationing on the government-bond issuance side, which, because the debt management office's objective function is unobservable, is downgraded to a supply-side background observation rather than an identified link in the causal chain). The foundation Paper One lays for the entire chain is twofold. First, at the level of judgment: it establishes the first main thread running through the whole book — "what enters allocation is a refracted shadow rate, not the interest rate itself." Second, at the level of mechanism: at the settlement-physics layer it reveals that the hard constraint on allocation ultimately lies not at the preference level of "willingness" but at the physical layer of "whether settlement can occur, whether leverage can be added" — the same government bond is simultaneously an asset and collateral, procyclical margin models cause collateral to withdraw financing at the very moment its value falls, and cash variation margin re-prices duration hedging as implicit cash-shorting. This finding at the settlement layer is precisely the mechanistic precursor to the decumulation-phase spiral of Paper Four. As for the question Paper One repeatedly touches but cannot resolve within its own framework — why accounting bases differ so much across countries, and why the welfare incidence of de-risking differs so much — this is handed off to Paper Two. ### Paper Two, "Between Defined Benefit and Defined Contribution: The Political-Economic Origins of Pension Institutional Variation": DB/DC Are Corner Solutions on a Continuum, and De-Risking Is Intergenerational Redistribution Paper Two takes up the question left open by Paper One: where do differences in accounting basis come from? Its answer is to reconstruct the seemingly discrete DB/DC institutional dichotomy as a continuum parameterized by "degree of risk-sharing s" and "distributional kernel φ" — s equal to zero is the textbook DC case of pure individual bearing, s equal to one is the textbook DB case of full cross-generational or sponsor bearing, and real-world institutions almost never sit at either corner; their position is set by social-partner negotiation, and the buffer ceiling, smoothing window, and intergenerational weights are themselves political stakes. Paper Two's sharpest and most defensible advance is a normative insight: the collective buffer is a contested asset without clear property rights, so that every move along this continuum toward "de-risking," in welfare terms, is never a pure reduction in risk but rather a redistribution of wealth between generations, between active members and retirees — a wholesale transition systematically raises the account value of members present at the time (especially those near retirement with high political organization), while correspondingly shrinking the share left for absent generations. This judgment re-politicizes "de-risking" from a technical maneuver into a distributive act, directly echoing the "distributive non-payment" observed in public DB plans in Paper One, and elevates it from an isolated case to a general mechanism running throughout the institutional spectrum. At the other end of the principal-agent chain, Paper Two offers an equally counterintuitive mechanism: the sign of allocation under DB de-risking and DC aggressiveness at low interest rates is driven by the option position of whichever party is the writer of this "put option" of residual guarantee, and this position can be orthogonal to, or even opposite from, that entity's nominal risk preference. Risk is pushed along the chain "sponsor—trustee—beneficiary—taxpayer—absent generations" to different resting points, and the hardness of the put option at the resting point determines the direction of allocation: the harder the guarantee (a strong employer, an enforceable contract), the more the guarantee-holder is motivated to de-risk in order to protect the contract — so the higher the funding ratio, the more de-risking occurs; the softer the guarantee (a weak employer, a public backstop, a collectively pooled intergenerational option), the more the put-option holder is motivated by upside — so risk is retained or even increased. This sign reversal of "the harder the guarantee, the more de-risking" is something a textbook risk-preference framework cannot predict at all, and it is the key that allows Paper Two to reconstruct the sign of allocation from a preference question into an option-structure question. On the DC side there likewise exists a "legally-chosen soft put option" — safe-harbor liability relief rewrites individually borne risk as institutionally borne risk chosen by law, thereby pushing up the equity center of gravity of the default glide path — which makes "who bears the risk that has been transferred away" the true object of Paper Two's political-economic analysis. On honesty, Paper Two makes a key self-convergence. It distinguishes two directions: Direction A (institutional form given, identify its causal effect on allocation behavior) is an identifiable skeleton — how the hardness of the writer of the residual-guarantee "put option" determines the sign of allocation (harder guarantees paradoxically produce more de-risking, a falsifiable and counterintuitive sign), how the non-interruptibility of mandatory contribution flows manufactures a "synthetic long-duration asset" on the asset side, and how the stringency of recognition rules drives de-risking; Direction B (allocation pressure reshaping the institution in reverse) has no clean identification anywhere in the paper and is honestly positioned as a theoretical proposition and political-economic case narrative. This convergence is not a retreat but a way of cutting the grand question of "why institutions are the way they are" into two halves: an identified consequence-side mechanism and a theorized origin-side narrative. What Paper Two lays as groundwork for Paper Three is a deeper question. Given that institutional form is continuously adjustable and de-risking is a distributive act, whether an institution can carry deeper alternative-asset allocation, whether it can make its assets as "advanced" as Canada's, is no longer a question answerable by the anthem of "holding alternatives equals being advanced" — it depends on whether the institution possesses the corresponding governance capacity, and the nature of that governance capacity is precisely the crux of Paper Three. The lens of "institutional complementarity" introduced at the end of Paper Two (pensions are embedded in the larger configuration of varieties of capitalism, financial system, and skill regime, and the direction of vulnerability is a property of configuration rather than an inherent property of institutional type) also becomes directly the theoretical gate for Paper Five's judgment of "non-transferability." ### Paper Three, "Alternative Assets, Global Allocation, and the Endogeneity of Governance Capacity": Capability Can Be Purchased Externally, Governance Control Rights Cannot Be Transferred Paper Three takes up the question left by Paper Two — whether an institution can deepen its allocation depends on governance capacity — and immediately performs a deeper cut on this correct but vague notion of "governance capacity." Its core judgment is: capabilities such as investing, valuation, and risk control are intermediate goods that can be priced, bought, and hired in factor markets, and therefore do not constitute a non-transferable moat; what is truly non-transferable is the allocation of **control rights** over capability — who owns deal origination, who holds veto power, whose investment mandate cannot be politically revoked. What can be copied is precisely capability; what cannot be copied is control. This judgment nails the answer to the technical question of "can Canada be emulated" down to: "capability can be learned, control rights cannot." Along this main line, Paper Three performs a four-layer progressive dissection, each layer negating a popular "advanced-practice" narrative. First, it exposes that a large portion of the "true premium" exhibited by the Canadian model is a statistical illusion manufactured by valuation smoothing — private-market valuation smoothing suppresses realized volatility and correlation, systematically inflating the Sharpe ratio and diversification metrics; moreover, book-value smoothing carries genuine demand-side value for weaker-governance plans (helping them "buy the right not to see volatility" in order to pass funding-ratio tests), with the speed of error-correction rising with governance quality. Second, it points out that the stronger the governance capacity and the faster the allocation, the faster the forward-looking premium is drained by its own reflexivity, so that capability is an accelerator rather than a moat — GPIF-style first movers capture a one-time competitive rent, while later replicators obtain only the post-competition beta. Third, it descends to the most difficult layer, that of private credit instruments, pointing out that what pensions buy is not credit but "non-default" as an actively manufactured product: through payment-in-kind conversions, debt-management actions, and rollover timing, default is hidden in a "forbearance inventory," whose truth is revealed only at fund liquidation when final realized losses become visible. Fourth, it repositions the true constraint on capacity from scale to the hardness of liabilities and the procyclicality of patient capital — hard, predictable DB liabilities constitute a "commitment device" that locks in funds and strips away opportunistic redemption rights, thereby actually enabling deeper illiquid allocation; sovereign wealth funds' "liability-free" status means funds can be withdrawn at any time, and the absence of commitment instead compresses illiquid capacity. These four layers of dissection jointly also touch on a device institutions try to use to compensate for insufficient governance — the proliferation of fairness opinions and compliance certificates — and reverse it: mandated fairness opinions do not supplement genuine judgment but instead substitute compliance certificates for genuine scrutiny, diluting accountability, constituting a governance-version of Gresham's law where bad practice drives out good. The identification design for this reversal was originally clean (the effective date of the relevant rule is exogenous to any single transaction), but since the regulation that served as its exogenous source was wholesale vacated by the courts in mid-2024, causing the event window to collapse, Paper Three honestly downgrades this leg's label from "identified" and retreats to "identification design in place, and the problem domain and market scale are verifiable." At the same time, Paper Three also downgrades the Canadian-style narrative that "internalization equals capability advantage": more than half of the net-return advantage of internalization comes from the saved fee wedge and the reallocation of performance compensation, rather than from stronger credit capability — capability here is a sufficient rather than a necessary condition — a judgment tested by re-estimating a shadow fee for internalized programs and then examining whether the residual, after stripping out the fee wedge, rises with a proxy for control rights, thereby cleanly resolving the attribution dispute over "internalization advantage." Paper Three's most important structural honesty is that, after rigorous self-examination, it acknowledges: the asset-side "ceiling on control rights" and the liability-side "activation of commitment" are two projections of **the same** underlying variable of political tolerance (extractability) onto the two sides, rather than two independent dual foundations. The original formulation that "asset-side control rights and liability-side commitment constitute an orthogonal duality" is honestly recognized as double-counting the same variable to artificially inflate the density of judgments; after convergence it becomes a stronger and more unified framework — political tolerance is the single endogenous axis running through both the asset and liability sides, manifesting on the asset side as a ceiling on control rights, and on the liability side as opportunistic redemption rights being politically reactivated. Paper Three thereby lays the interface for Paper Four: "homogeneously held alternative assets undergo synchronized liquidity evaporation under stress" — earning the premium (an asset-side benefit) and bearing liquidity risk (a liability-side risk) are in fact two sides of the same mechanism, and the risk side of this coin is precisely the theme of the decumulation phase. In addition, Paper Three honestly downgrades several sharp judgments to theoretical propositions with attached falsification conditions — "non-transferability of control rights" requires first establishing an operationalized proxy that precedes performance and can be independently coded, failing which the associated argument concedes retreat to a taxonomic framework; "manufacturing non-default" must be made visible via the ex post truth event of final realized losses at liquidation, failing which it is not presented as an identified mechanism — this practice of separately labeling sharp judgments and their identification burden is a microcosm of the whole paper's stance of honesty. ### Paper Four, "The Endogeneity of Risk Governance and Accountability in the Decumulation Phase": Accountability Is an Endogenous Institutional Choice, Buffers Only Shift the Trigger Point, and the Non-Explosion of the Dutch/Danish Systems Is Survivorship Bias Paper Four takes up the interface left by Paper Three — homogeneous-holding liquidity risk — and shifts the entire analysis to the decumulation phase, a new window of stress. Its spinal judgment is: the interest-rate decumulation phase does not add a new source of risk, but rather, within a pre-bindable mapping of "institutional variable → accountability dimension," exposes each country's endogenous institutional choice of accountability. Paper Four is a corrective work, sequentially correcting five progressively deeper misdiagnoses. First, within the UK it falsifies the optimistic signal that "improved funding ratio equals safety": a rise in interest rates is a two-sided event for DB with an undetermined net-effect sign, simultaneously depressing both the present value of liabilities and the value of collateral, so that funding-ratio improvement and margin calls are jointly driven by interest rates, and the former carries no informational value for the latter — or even carries the opposite — improvement in the funding line does not constitute a safety signal for the liquidity line. Second, it uses survivorship bias as a "symmetric knife": if the survival of the Netherlands and Denmark is in fact sub-critical luck, then this data provides no information at all about "whether the institution is superior" — it cuts down both the inference of "Dutch/Danish superiority" and the reverse inference of "the UK is not special"; what is genuinely identifiable is using the pseudo-registration locations of Ireland and Luxembourg as controls to falsify the jurisdictional dichotomy of "the UK equals risk, the Continent equals safety," as well as treating the UK's internal vehicle structure (highly leveraged pooled vehicles versus lower-leveraged segregated vehicles) as the cleanest quasi-natural control standing on its own. Third, it precisely locates the pathology of the spiral in the mismatch between the monetary nature of margin and the rescuability of the central bank — the spiral of synthetic duration is welded, via clearinghouse cash margin, into a cash spiral that cannot self-heal, and rescue on the securities side is ineffective on the cash side — but honestly positions this diagnosis as a theoretical-dynamics proposition whose empirical identification awaits external scope variation. Fourth, it reconstructs the independent variable of stress testing from a static basis-point endpoint into a functional of path, density, and correlation structure (rarity should be defined by a minimum-density domain rather than integer basis-point shifts), which is a defensible methodological-level contribution, while "path matters more than magnitude as the cause of death in 2022" is honestly downgraded to an empirical hypothesis awaiting multi-event panel testing. Fifth, and the meta-judgment of the whole paper: the location of risk is an endogenous institutional choice of accountability, thickening buffers merely shifts the trigger point rightward rather than structurally eliminating it, and endogenizing central-bank backstops changes the topology of the next round of contagion — but this highest-level judgment is narrowed from an unfalsifiable universal claim of "everything is endogenous" to a comparative proposition that allows "successful transplantation" as a falsifier. Paper Four's methodological discipline is especially worth pointing out, because it pushes the stance of "honest downgrade" to its most thorough. It repeatedly distinguishes "availability of method" from "identifiability of proposition": the method of minimum-density domains is mature, and the fact that the 2022 correlation structure flipped from negative to positive is measurable, but the claims that "path is causally more important than magnitude" and "the correlation flip is the true culprit this time" mistake correlation for causation within a single event, unable to be distinguished from "the interest-rate shock simultaneously drove both the correlation flip and the crash" or "the correlation flip is endogenous to the sell-off," and can therefore only stand as untested hypotheses. It also rejects "outcome-defined" circularity — if "overallocation that should have held up but was wrongly killed" can only be defined by the ex post price rebound, then it is unfalsifiably true by construction, so it demands an ex ante computable "should-retain set" criterion based on liquidity premium, fundamental quality, and spiral correlation, and requires an out-of-sample comparison using ex ante rules rather than ex post labeling of "wrongly killed." This practice of narrowing unidentifiable, unfalsifiable claims one by one occurs a full nine times within Paper Four, which makes this corrective work itself withstand correction. What Paper Four lays as groundwork for Paper Five is a crucial premise: given that accountability is endogenous, that buffers only shift the trigger point, and that best practice cannot be simply transplanted, any late-developing economy designing its own decumulation-phase risk governance cannot hope to transplant a ready-made scheme from mature pension markets, but must design endogenously. This is precisely the starting point of Paper Five in addressing the question of China. ### Paper Five, "Adapting China's Multi-Pillar Pension System to Low Interest Rates": The Five Pillars Are Incommensurable and Cannot Be Copied Wholesale Paper Five brings the entire matrix to its conclusion, bringing the diagnoses of the preceding four papers to bear on China's multi-pillar pension system. Its spinal judgment is: the differences among the five pillars are, in essence, layered misalignments in "who holds the property right to which category of risk, and has the capacity — not merely the obligation — to price and provision for it"; the low-rate environment is not an external shock but rather renders this pre-existing misalignment visible, turning it from implicit to explicit. "Incommensurability" here is not an excuse to refuse comparison — the five pillars can be explicitly ranked and compared along four commensurable dimensions (fund maturity, liability duration, the right to define the risk-free anchor, and the recipient entity at decumulation) — it refers only to the bounded proposition that "the same allocation logic cannot be applied uniformly across pillars." Paper Five closes the loop with the preceding four papers in five progressive steps. At the governance level (closing the loop with Paper Three), it points out that the governance superiority of the National Council for Social Security Fund is not a transplantable best practice but a one-time "grant" of stock endowment arising from its constitutional position close to the center; governance quality drops in discrete steps across the three tiers of "close-to-center, provincial, and local." Here Paper Five makes an honest self-restraint — rather than writing this gap as a "distance law" of independence continuously decaying with distance from the center (three discrete tiers cannot identify a continuous function, and distance and incentive are nearly perfectly collinear in China, constituting a spurious opposition), it converges to a discrete proposition of stepwise decline across three tiers, which is the strongest claim that the cross-section of the three-trustee structure can support. Meanwhile, the top-down attempt to synthesize patient capital via "sales bans plus long-term holding plus mandatory capital injection" in fact retains the obligation while stripping away the right to rebalance and cut losses, manufacturing an institutional reverse-rebalancing of "the lower the return, the more is added" — retaining obligation is not equivalent to patient capital, and this "self-destruction of property rights in synthesizing patient capital" can be tested using the 2024 use-restriction measures as an institutional break, via an event study on the same fund (whether the absence of rebalancing discipline under sales-ban constraints raises left-tail risk). On the asset and fiscal axis (closing the loop with Paper One), it reveals that there fundamentally exists no exogenous "risk-free anchor" on the asset side of Chinese pensions: it is a target endogenously moved by fiscal authorities — the first and second pillars are institutionalized as the "final buyer of last resort" for debt resolution, the "phantom account" of occupational annuities is a contingent pension liability recorded on the fiscal books, and the accounting basis of "seven percent return, actuarially neutral" systematically overstates true returns, with the positive wedge being an off-the-books implicit intergenerational tax — this inverts the "stable risk-free anchor" assumed by Paper One, and is a mechanism unique to China. At the liability-pricing level (closing the loop with the visibility theme of Paper Four), it points out that the allocation bifurcation of the second pillar is driven by accounting visibility rules rather than genuine changes in cash flow — the same contribution is seen through two different risk-free rates under dual bookkeeping rates, and trustees treat the bookkeeping rate as an implicit opportunity-cost anchor and contract defensively. At the behavioral level (closing the loop with Papers One and Two), it reconstructs the "cold" contribution behavior of the third pillar from an issue of individual literacy into an issue of institutional regressivity — uniform nudges are regressive for low- and middle-income individuals, closed accounts are an unpriced prepayment of a liquidity option, and what is truly transferable is the default mechanism rather than the rate of return, and the transferability of the default mechanism itself depends on whether a supporting institutional carrier exists. At the decumulation end (closing the loop with the risk boundary of Paper Four), it points out that there is a misalignment and soft constraint in the entity receiving longevity risk — the payout end does not mandate annuitization, so that the longevity risk of the collective pool is disaggregated household by household and returned unchanged to families (a reverse socialization of responsibility); for those who are neither annuitized nor covered by the first pillar, families are also unable to bear this risk, which ultimately settles as an unpriced, unprovisioned soft fiscal liability. What Paper Five provides for the entire matrix is a unified criterion of "transferability": an overseas mechanism is transferable if and only if the institutional carrier on which its effectiveness depends (a governance carrier, a default carrier, an exogenous risk-free anchor, a decumulation-end recipient entity) is independently obtainable at the target pillar. Transferability is thereby transformed from a matter of the analyst's discretion into a verifiable proposition about the availability of preconditions. This criterion unifies all of the "cannot be simply copied" conclusions of the preceding four papers under a single verifiable standard, thereby drawing a conclusion for the entire matrix that is both pioneering and honest. --- ## 3. The Governing Judgments: Three Main Threads as the System's Soul If progressive logic is the skeleton of the matrix, what truly makes the five papers a system rather than five independent diagnoses is three core judgments that run through them from beginning to end. They are not a simple summation of the five papers' conclusions, but the unfolding of a single intellectual stance across five different levels. **The first main thread: what enters the allocation function was never the "interest rate" itself, but a shadow rate refracted layer by layer through institutional form, accounting, and fiscal policy.** This thread is formally established in the first paper — four sets of misaligned accounting refract into four shadow rates, and allocation is driven by the opening and closing of the accounting wedge. It descends one level in the second paper: differences in accounting treatment are rooted in differences in institutional form (position on a continuous spectrum), and institutional form itself is a product of political negotiation. It is given a dynamic dimension in the third paper: the so-called "true premium" is itself refracted by valuation smoothing, and there exists a governance-determined wedge between the volatility seen on the books and true economic volatility. It is flipped onto the risk side in the fourth paper: what the funding line shows and what the liquidity line demands are two different signals refracted by accounting and institutional form, and improvement in the former carries no information about the latter. It reaches its sharpest form in the fifth paper: China's risk-free anchor is not exogenously given at all, but a target endogenously moved by fiscal policy, and the accounting (crediting) rate is the price of fiscal redistribution rather than a return on investment. From beginning to end, this thread negates the same naive assumption — that there exists an unmediated interest-rate scalar that can enter the allocation function directly. **The second main thread: the key variables are endogenous, and endogeneity occurs precisely where people are most inclined to treat them as exogenous.** This thread runs through the identification layer of all five papers. The first paper points out that writing allocation as a function of the interest rate is least tenable precisely during a prolonged low-rate period when "exogenous rates" would be assumed — population aging is a common cause driving both rates and allocation simultaneously, and pension funds' rigid demand in turn reflexively produces the low-rate environment itself. The second paper points out that the risk-bearing subject is not a one-time attribute given by institutional type, but an endogenous stopping point pushed out along the principal-agent chain by low rates, with the hardness of its embedded option dominating the sign of allocation. The third paper pushes this thread to its core: governance capacity is not an exogenous endowment, control rights are endogenous, and the seemingly clean causal claim that "liability hardness determines allocation depth" is in fact endogenous with governance as a common cause — institutions capable of sustaining hard-liability contracts are precisely those that already had strong governance and political protection; governance is the common cause determining both simultaneously. The fourth paper identifies the attribution of responsibility itself as an endogenous choice — the placement of risk is not a technical necessity but an institutionally endogenous political-economic allocation. The fifth paper is a manifestation of endogeneity at every turn: governance independence is a stock endogenously granted by constitutional position, the risk-free anchor is a target endogenously moved by fiscal policy, contribution-rate coldness is behavior endogenously driven by institutional regressivity, and the misalignment in bearing longevity risk is a soft liability endogenously deposited by institutional migration. This thread has a shared methodological consequence: any identification strategy that treats an endogenous variable as an exogenous instrument, however tempting, is exposed and honestly downgraded one by one in this study (the repeated abandonment of demographic-structure instrumental variables is the most consistent methodological manifestation of this thread). **The third main thread: advanced-economy experience cannot simply be transplanted, because what is transferable is capacity, while the institutional preconditions that carry that capacity are not transferable.** This thread is the inevitable real-world destination of the first two threads. If the rate that enters allocation is institutionally refracted, and if the key variables are endogenous to institutions, then any prescription claiming to "replicate the best practice of country X" is bound to fail, because the transplanted action is severed from the institutional carrier that refracts and endogenizes it. The third paper supplies the core mechanism of this thread — capacity can be purchased externally, control rights cannot be transferred; copying Canada's homework gets you the capacity but not the control rights. The fourth paper supplies its form in the decumulation (payout) phase — attribution of responsibility is endogenous, buffers merely displace the trigger point, best practices cannot be transplanted, and the fact that the Netherlands and Denmark have not yet exploded cannot be used to infer institutional superiority. The fifth paper forges it into a single, checkable, unified criterion — transferability is equivalent to the independent obtainability of the institutional carrier at the target level, thereby foreclosing the possibility that "cannot simply be transplanted" degenerates into a shield against comparison. This thread is the real-world soul of the entire matrix precisely because it converts the seemingly purely theoretical diagnoses of the first four papers into a direct intervention in an actually ongoing policy process — pension reform in late-developing economies and the project of "learning from international experience." The three main threads interlock with one another: the shadow rate (thread one) exists precisely because the accounting and institutional forms that refract it are endogenous (thread two); and precisely because both refraction and endogeneity are rooted in a non-transferable institutional carrier, advanced-economy experience cannot simply be transplanted (thread three). The converse also holds — if the interest rate could enter allocation unmediated, there would be no bifurcation into a shadow rate, no question of "whether the institution refracting it is endogenous," still less any question of "whether the carrier is transferable," and all three threads would collapse simultaneously into the simple world envisioned by the mainstream dichotomy. Precisely because the three threads are mutual preconditions of one another, none dispensable, this study must unfold them as a single whole — and unfolding that whole requires passing in sequence through the five levels of accounting treatment, institutional form, governance, decumulation phase, and cross-country comparison. Together, these three threads answer the question posed at the outset of the general introduction — why five papers are needed: not because there are many topics, but because these three interlocking threads must each be argued through fully at every level; missing even one breaks the chain, leaving the main threads without one of their intermediate links and unable to substantiate themselves. --- ## 4. Methodological Self-Awareness: Judgment First, Logic as Core, Honest Downgrading This study maintains a consistent methodological self-awareness throughout, which determines both how the five papers are written and where their boundaries lie. This self-awareness can be summarized in five points. **First, judgment takes priority over statistical data.** The core of an economics paper is judgment, not the numb piling-up of data. Every paper in this study takes as its spine a falsifiable, directional core judgment, with data serving to test the judgment rather than to induce it. A direct consequence of this orientation is that this study rejects the common but pedestrian path of "systematically compiling data from several countries and inducing patterns from the data," instead insisting on first designing an original judgment and then letting that judgment submit to the verdict of evidence. Every section's argument across the five papers has, as its heading, a core judgment that can itself be refuted, rather than a neutral topic label; readers can tell from the table of contents alone that what this study presents is a set of assertions awaiting testing, not a classified literature review. **Second, design before implementation, slow modeling.** This study insists on thinking the design approach through thoroughly and originally at a high level before moving into implementation, rather than engaging in undisciplined, haphazard data exploration. This means that each paper's identification strategy, falsification threshold, and comparative design are carefully constructed only after the judgment has been established, rather than stumbled upon by chance within the data. This orientation gives the argumentative skeleton of the five papers a sense of "design" — the responsibility of each leg, the opposing predictions of each cross-section, and the rationale for each downgrade are all thought through in advance. **Third, logic is the core; language merely clears the pass line.** The core criterion of this study is logic, not ornamentation. Originality is first and foremost originality of judgment and logic, not stylistic flourish. What the five papers pursue is a genuinely proactive, original line of exploration; the language need only meet the stylistic baseline of a doctoral dissertation — it earns no extra credit for literary flair, nor is any vagueness of expression tolerated in exchange for sharpness of logic. **Fourth, honest boundaries take priority over bluster.** This is the most distinctive, and also the most unconventional, methodological self-awareness in this study. At the level of identification, much of the data required for micro-level identification — plan-level collateral architecture, a cross-nationally comparable index of ALM stringency, realized losses at liquidation of private-equity funds, the distribution of contribution density among flexibly employed workers — is, under the conditions of this study, often unobtainable. Faced with this reality, this study chooses explicit downgrading rather than fabrication: any judgment that cannot be pinned down is uniformly and explicitly labeled a "mechanism hypothesis," "case study," "theoretical proposition," or "hypothesis awaiting pre-registration," rather than dressed in identification-style language to package an assertion that is in fact unidentifiable. The first paper downgrades the distributive politics of public DB plans to a qualitative case study (because it is trapped within an accounting identity) and downgrades duration rationing on the issuance side to a supply-side background observation; the second paper downgrades all assertions concerning Direction B to theory and case study; the third paper downgrades key propositions such as "control rights are non-transferable," "collusive claw-back," and "manufactured non-default" to theoretical propositions with attached falsification conditions; the fourth paper honestly downgrades nine claims that are unidentifiable or unfalsifiable; and the fifth paper draws a hard line — the absence of evidence only weakens the strength of a claim and can never be reversed into positive evidence for the opposing hypothesis (treating "no trace found" as proof that "the trace does not exist" is explicitly prohibited). This honest downgrading is not a failure of argument — quite the opposite, it is a quality upgrade that converts an empty check into a defensible assertion: a sharp judgment honestly labeled a "theoretical proposition" has greater academic value than an unfalsifiable empty claim disguised in identification-style rhetoric. **Fifth, self-awareness of dissertation conventions.** In finalizing the text, this study follows the conventions of a doctoral dissertation — academic written register, in-text citations and a reference list, and standardized figures and tables — while ensuring zero leakage of language (leaving no trace of procedural or instrumental phrasing). This stylistic self-awareness does not conflict with logic being the core: logic remains the core, and meeting the stylistic standard is a necessary condition for the finished product, not a bonus. These five points of methodological self-awareness are not mutually isolated; together they serve a single conviction: the weight of a good economics paper comes from the originality of its judgment and the rigor of its logic, not from the accumulation of data or the ornamentation of language — and for the originality of a judgment to truly stand, it must be paid for with honest boundaries. Better to label a sharp judgment "a theoretical proposition awaiting micro-level data" than to dress up an unfalsifiable empty claim in identification-style rhetoric. It is precisely this conviction that gives the five papers such high methodological unity: all take a falsifiable core judgment as their spine, all design before implementing, all explicitly downgrade where identification is impossible, and all pursue dissertation conventions in the finished text. This methodological unity, together with the interlocking of the three main threads in content described above, constitutes the twin guarantees that make this study a system rather than a patchwork — interlocking threads in content, consistency of method throughout. --- ## 5. Contributions and Limitations of the System **In terms of contribution**, the originality of this study relative to the existing literature is concentrated in its wholesale rejection of the mainstream "reaching for yield vs. ALM" dichotomy, and in the three interlocking main threads that unfold from this rejection. It reconstructs the independent variable of the allocation function from an "interest-rate scalar" into a "shadow-rate vector refracted by institutional accounting"; it reconstructs the discrete DB/DC dichotomy into a continuous spectrum of risk-sharing degree, and re-politicizes de-risking as intergenerational redistribution; it shifts the moat of governance capacity from "capacity" to "non-transferable control rights," exposing numerous statistical illusions in advanced-economy experience; it flips the governance of risk in the decumulation phase from "improvement means greater safety" to "the endogenous political economy of attributing responsibility," proving that buffers only displace the trigger point; and it finally distills all of this into a framework for China's five-layer misalignment in bearing risk, together with a checkable criterion of transferability. Each of these five reconstructions directly counters a popular policy narrative — "allocating to alternatives means being advanced," "learn Canada's governance," "replicate the Dutch hedge," "thicken the buffer and you are safe," "just perfect the multi-pillar pension system" — and their combined force constitutes a judgment-driven diagnostic framework for the systemic reconstruction of pension allocation under low interest rates and in the decumulation phase. **In terms of limitations**, this study maintains full self-awareness. First, and most importantly, most of the identification in this study belongs to the design layer rather than the completed empirical layer — for the great majority of the core judgments across the five papers, the identification strategy has been carefully designed, but truly clean exogenous shocks and usable micro-level data cover only a small portion of them (for instance, the standards-differencing in the first paper and the phased liquidation obligations in the fourth paper provide relatively exogenous identification timing points, but even for this small portion, the micro-level coefficients remain largely unestimated under current conditions — only the identification design and the timing itself are verifiable; the regulatory discontinuity that the third paper originally relied upon has, moreover, retreated to being merely an identification design after its exogenous source was wholesale overturned by judicial action). The bulk of the remaining judgments await the filling-in of micro-level data before they can be upgraded from "theoretical proposition" or "hypothesis awaiting pre-registration" to "identified." This study chooses to honestly label this state rather than conceal it. Second, several key comparable indicators (cross-national ALM stringency, institutional complementarity, realized losses at liquidation of private equity, the distribution of contribution density) are, under current conditions, missing or require extensive institutional verification to construct — this constitutes a shared bottleneck for identification in this study. Third, the sample window for the decumulation phase remains short (only a few years since 2022), and the identifying power of several auxiliary tests that rely on sign reversals in the decumulation phase is constrained by sample size — this study faithfully notes this limitation. Finally, for the analysis concerning China, many of the mechanisms (the scale of debt-resolution absorption, the absolute level of the accounting/crediting-rate wedge, the amount of fiscal soft liabilities absorbed) can, at present, only be given as directionally falsifiable sign judgments; their absolute levels await further data. These limitations do not diminish the contribution of this study — rather, they define its nature: this study is a judgment-driven, design-complete, honestly bounded diagnostic framework. It provides a set of mutually interlocking, original judgments and executable identification designs for a major real-world problem still unfolding, while clearly marking which judgments are already borne out by evidence and which await final adjudication by future micro-level data. In a field where the data are not yet complete but the problem is already urgent, making the judgment and the design as original and as honest as possible is the most responsible stance this study has chosen — and the one it is able to contribute.