In most non-life insurance carriers, pricing and reserving are treated as distinct actuarial functions that happen to use some of the same underlying data. The pricing actuary builds loss development assumptions to project ultimate losses for next year's business; the reserving actuary builds loss development assumptions to project ultimate losses for prior accident years in force. The assumptions are related. The datasets overlap substantially. In many smaller carriers, the same actuary does both. But the models are typically maintained separately, the calculations run at different points in the quarter, and the assumption sets are reconciled manually, if they are reconciled at all.
This separation is partly historical. Pricing and reserving actuarial certifications, regulatory reporting structures, and professional standards have traditionally treated the two functions as addressing different questions: reserving answers "what is the carrier's current liability for in-force policies and claims" while pricing answers "what premium is adequate for next year's expected losses." The questions are different, but the empirical base they draw from is the same loss triangle data, and the development pattern assumptions that feed both analyses should, in principle, be consistent.
In practice, they often are not, and the divergence creates problems that surface at awkward times.
Where the Assumptions Diverge and Why It Matters
A pricing actuary selecting loss development factors for a commercial auto line will work from a triangle that typically excludes the most recent accident years where development is still immature. The selection process considers the stability of the observed age-to-age factors, applies actuarial judgment to exclude outlier periods, and may use an industry development benchmark as a credibility supplement. The selected development factors form the backbone of the pure premium calculation that feeds the rate indication.
A reserving actuary working from the same underlying triangle, but applying the Bornhuetter-Ferguson method with the prior year loss ratio as the a priori estimate, may produce a slightly different development factor selection for the same line of business, because the BF method's credibility weighting changes the relative influence of the selected development factors versus the prior expected loss ratio. The selected factors may look the same on paper, but the way they interact with the prior estimate in the BF calculation can produce different implied ultimate loss patterns than the pricing model assumes.
This matters when the pricing actuary presents a rate indication showing expected loss ratio improving relative to the prior year reserve position, and the reserving actuary simultaneously presents a reserve that implies the prior year loss ratio will worsen. The two actuaries are using the same loss data, but their model architectures have produced contradictory signals about trend direction. In an organization where both analyses flow to the same CFO or CRO meeting, this creates a credibility problem that neither actuary can resolve without understanding how the other's model works.
The Assumption Reconciliation Problem in Practice
The standard industry practice for managing pricing-reserving assumption consistency is an annual or semi-annual reconciliation meeting where the two actuarial functions review their respective development factor selections and identify divergences. In large carriers with dedicated actuarial departments for each function, this is a structured process with a formal output document. In smaller carriers where a single team handles both functions, it is often an informal conversation that does not produce a documented record.
The problem with the reconciliation meeting approach is timing. The meeting happens after both sets of assumptions have been selected, after rate filings have potentially been submitted to regulators, and after reserves have potentially been signed off by the appointed actuary. At that point, a significant divergence in development factor selections between the two models requires either revising the pricing model assumptions (which may trigger a rate re-filing if the indication changes materially), revising the reserve assumptions (which affects the balance sheet), or concluding that the divergence is justified by methodological differences. The third option is often the path of least resistance, even when the divergence is not actually justified.
A model architecture where pricing and reserving calculations draw from the same assumption library, with deviations explicitly documented and justified at the time of selection rather than discovered and rationalized after the fact, eliminates this problem structurally. It does not eliminate the need for actuarial judgment; it ensures that the judgment is exercised once, consistently, with an explicit record of the rationale, rather than twice, separately, with reconciliation as an afterthought.
What a Shared Data and Assumption Layer Looks Like
The core of a combined pricing-reserving model cycle is a shared loss triangle that serves as the authoritative input for both functions. This triangle is extracted from the claims administration system, validated and reconciled at the start of each model cycle, and versioned so that both the pricing and reserving calculations use an identical copy of the same input data cut at the same date. When the triangle is updated for the next cycle, both models update together.
Above the shared data layer, the assumption library contains the development factor selections for each line of business, along with the selection rationale, the alternatives considered, and the sign-off. When the pricing actuary selects development factors for the next pricing model cycle, those selections are recorded in the assumption library. The reserving actuary reviewing the same factors for the IBNR calculation can see the pricing actuary's selection and either adopt it or document why the reserving application requires a different selection. The documentation creates an explicit record of any divergence and its justification, rather than leaving the divergence implicit in two separately maintained spreadsheets.
The tail factor, which is often the most significant source of pricing-reserving divergence in long-tail lines like commercial general liability or workers compensation, gets particular attention in a shared assumption framework. The pricing model's tail factor selection reflects the actuary's view of expected development beyond the observable triangle. The reserving model's tail factor selection for the same line reflects the actuary's view of the remaining development in the held reserve. These should be derived from the same empirical analysis of industry benchmarks and the carrier's own observed tail behavior. Maintaining them as separate values in separate models invites divergence that may not be detected until a reserve review or a rate hearing surfaces the inconsistency.
The Quarterly Cycle Timing Challenge
One practical objection to a tightly integrated pricing-reserving model cycle is that the two functions operate on different time scales. Statutory reserve calculations run quarterly, with a signed actuarial opinion required at year-end. Rate filings, depending on the line of business and the regulatory jurisdiction, may be filed annually or on a rolling basis as experience accumulates. The quarterly reserving cycle does not naturally synchronize with the annual or semi-annual pricing review cycle.
This is a real constraint, not a fabricated obstacle. A shared assumption framework does not require that pricing and reserving calculations run at exactly the same time. It requires that the development factor assumptions used by both are consistent with each other and that any differences are explicitly documented. The reserving team can run its quarterly calculation using the most recently reviewed and documented development factor assumptions without waiting for the annual pricing cycle. When the annual pricing cycle runs, it can draw from the most recently finalized reserving development factors, update where appropriate, and document the updates in the shared assumption library.
The key mechanism is that assumption changes in either direction are visible to both functions. When the reserving actuary updates the development factor selection for a line based on two quarters of adverse development, that update is recorded with a timestamp and the supporting data. The pricing actuary reviewing that update at the next annual pricing cycle has contemporaneous documentation of what changed, when, and why, rather than having to reconstruct the development from a comparison of two spreadsheet versions separated by six months.
Regulatory Implications: Rate Filings and Reserve Certifications
The practical regulatory benefit of a consistent pricing-reserving framework is most visible in jurisdictions where the same actuarial team supports both the appointed actuary's opinion on reserves and the supporting actuarial materials for rate filings. When a rate hearing or a regulatory examination asks how the pricing development factors were determined and whether they are consistent with the reserve development factors, the answer should not require cross-referencing two independently maintained spreadsheets and explaining divergences that were never formally reconciled.
A rate filing actuary who can point to a single development factor assumption record, with documented rationale and an explicit notation of where the pricing application differs from the reserving application and why, is in a substantially stronger position than one who must explain a divergence that appears inconsistent but was never formally addressed. The regulatory reviewers reviewing rate adequacy and the financial examiners reviewing reserve adequacy may not coordinate their reviews, but the documentation they each receive should tell a consistent story about the carrier's actuarial judgment on loss development.
We are not suggesting that pricing and reserving should always use identical development factors. There are legitimate methodological reasons why the two applications may differ, and those reasons should be documented. What we are saying is that a framework that forces that documentation to happen at the time of assumption selection, rather than in retrospect when an inconsistency is discovered, produces better outcomes for the carrier and cleaner documentation for regulators.
What Changes in Practice
Moving to a combined pricing-reserving model cycle does not require restructuring the actuarial department or changing the professional responsibilities of the pricing and reserving actuaries. It requires three things: a shared, versioned data layer; an assumption library that serves both functions with explicit cross-function visibility; and a review workflow that includes a documented check for inter-function consistency before either set of assumptions is finalized.
The review workflow addition is the most significant behavioral change. Currently, the pricing actuary finalizes development factor assumptions before or after the reserving actuary, depending on the timing of their respective work cycles, and neither necessarily reviews the other's selections before committing to them. In a combined cycle, a consistency review is a required step before finalization of either, with the output being either documented agreement or documented justification for the difference. This adds perhaps two to four hours to the cycle for lines where divergence requires discussion; it is close to zero additional effort for lines where the selections naturally converge.
The output is a cleaner actuarial record, a more defensible regulatory position, and the elimination of a category of embarrassing inconsistency that surfaces at the worst possible moments: during a reserve review following adverse development, or during a rate filing hearing when a regulator notices that the filed loss development pattern contradicts the loss development pattern implied by the statutory reserve.
