Tutorials › CAS Exam 8 - Advanced Ratemaking › Applying Customer-Level Price Movement Guardrails

CAS Exam 8 · Optimization Implementation and Testing · Tutorial 869 of 989

Applying Customer-Level Price Movement Guardrails

Apply quote-level guardrails to candidate renewal prices and determine whether an allowed recommendation exists.

Advanced 9 min read

What You'll Learn

  • Convert percentage and dollar movement rules into allowable premium ranges
  • Combine price limits with policy and quote eligibility conditions
  • Filter a discrete price menu before selecting its best recommendation
  • Identify empty feasible sets and explain why a recommendation cannot be issued
  • Avoid common errors in movement calculations and guardrail application

Why Customer-Level Guardrails Matter

A price optimization process may identify a candidate renewal price with attractive expected contribution, but that candidate is not automatically an acceptable quote. A customer-level guardrail can restrict how far the premium may move, prohibit a quote for a particular policy condition, or impose a minimum or maximum premium. The recommendation must satisfy all applicable conditions at the same time.

This tutorial focuses on applying those conditions to the discrete candidate prices available for an individual customer. It does not replace the underlying price or demand analysis: it filters the candidates that analysis supplies. The result can be the best allowed candidate, or an empty feasible set in which no candidate meets every rule. This implementation discipline is aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus.

Definition: Let \(P_0\) be the customer’s current premium on the stated comparison basis, and let \(P_j\) be candidate renewal premium \(j\). A customer-level price movement guardrail is a rule that a candidate must satisfy for that customer before it may be recommended. A candidate is feasible only if it satisfies every applicable movement, eligibility, floor, and ceiling rule.

Translate Movement Rules Into Price Bounds

A percentage movement is measured relative to a specified base premium. For a renewal comparison, define the signed percentage change for candidate \(j\) as

$$ r_j=\frac{P_j-P_0}{P_0}. $$

A negative value is a decrease; a positive value is an increase. If the permitted movement is from \(g\) to \(h\), where \(g\leq h\), then a candidate must satisfy \(g\leq r_j\leq h\). Assuming \(P_0>0\), multiply by \(P_0\) and rearrange to obtain the allowable premium interval:

$$ g\leq \frac{P_j-P_0}{P_0}\leq h \quad\Longleftrightarrow\quad P_0(1+g)\leq P_j\leq P_0(1+h). $$

This is the direct conversion from percentage-change language to dollar bounds. For example, a permitted movement of -5% to +12% means the candidate may be as low as 95% of the current premium and as high as 112% of it. By contrast, if the rule is stated as a dollar decrease of at most \(D_-\) and a dollar increase of at most \(D_+\), its bounds are \(P_0-D_-\) and \(P_0+D_+\). Do not treat a dollar limit as a percentage limit or vice versa.

The comparison basis must be consistent. If the guardrail applies to premium before fees, compare before-fee amounts; if it applies to a stated total, use that total for both the current and candidate prices. Do not mix bases. The examples below treat all stated premiums as being on the same comparison basis.

Feasibility With a Discrete Price Menu

A discrete menu is a finite set of candidate prices already generated for the customer. The feasible set is the subset that survives all customer-level rules. A menu entry can fail even when its expected contribution is high: expected contribution ranks acceptable candidates, but it does not make a prohibited candidate acceptable.

Definition: For a menu of candidates \(j=1,\ldots,n\), let \(E_j\) equal 1 when candidate \(j\) is eligible and 0 otherwise. Let \(L\) and \(U\) be the combined lower and upper premium bounds. The feasible set is $$ \mathcal{F}=\{j:E_j=1,\ L\leq P_j\leq U\}. $$ If a candidate has multiple eligibility conditions, \(E_j=1\) only when all of them are met. If \(\mathcal{F}\) is nonempty, select the best candidate under the specified objective from within \(\mathcal{F}\). If \(\mathcal{F}\) is empty, there is no feasible recommendation under the stated rules.

For example, a percentage movement interval, a minimum premium, and a maximum premium can be combined by taking the greatest of the lower bounds and the least of the upper bounds. Eligibility is then checked for each menu candidate. A premium inside the combined numeric interval is still infeasible if an eligibility condition fails.

1
Fix the comparison basis.
Record the current premium and make sure candidate prices and guardrail amounts refer to the same premium basis.
2
Convert every numeric rule.
Translate percentage and dollar movement limits into lower and upper premium bounds. Include any applicable premium floor or ceiling.
3
Test each menu candidate.
Check its price against the combined bounds and verify every required eligibility condition.
4
Choose or report infeasibility.
Rank only feasible candidates using the stated objective. If none survive, report that the menu has no feasible recommendation under the rules.

Worked Applications

Worked Example: Percentage Guardrails and a High-Value Candidate

A hypothetical insurer is evaluating four renewal prices for a policy with a current premium of $800. The permitted movement is from -5% to +12%. The candidate premiums and their modeled expected contributions are: $750 with $118, $780 with $126, $840 with $132, and $910 with $149. Which price should be recommended? Use expected contribution as the objective, but only among feasible prices.

Step 1: Convert the movement limits. The lower bound is \(800(1-0.05)=800(0.95)=\$760\). The upper bound is \(800(1+0.12)=800(1.12)=\$896\). Thus the allowed premium interval is [$760, $896].

Step 2: Test each candidate. For $750, the change is \((750-800)/800=-50/800=-0.0625=-6.25\%\), so it is below the -5% limit. For $780, the change is \(-20/800=-2.50\%\), so it is allowed. For $840, the change is \(40/800=5.00\%\), so it is allowed. For $910, the change is \(110/800=13.75\%\), so it exceeds the +12% limit.

Step 3: Optimize over the feasible candidates. Only $780 and $840 are feasible. Their expected contributions are $126 and $132, respectively. The best feasible candidate is therefore $840, with expected contribution $132.

Check. $840 lies between $760 and $896, and its 5.00% increase lies between -5% and +12%. The infeasible $910 quote has the highest contribution in the menu, but it cannot be selected under the stated guardrail. The method is to filter first and rank second; this avoids allowing an objective score to override a hard rule.

Worked Example: Movement Limits Plus an Eligibility Ceiling

A hypothetical renewal has a current premium of $960. The movement rule permits a decrease of at most 6% and an increase of at most 9%. A separate quote-level eligibility rule for this policy allows only candidates at or below $1,010. The menu is $900, $930, $1,000, and $1,040, with modeled expected contributions of $54, $68, $72, and $79, respectively. Find the best feasible candidate.

Step 1: Calculate the movement interval. The lower movement bound is \(960(1-0.06)=960(0.94)=\$902.40\). The upper movement bound is \(960(1+0.09)=960(1.09)=\$1,046.40\).

Step 2: Apply the eligibility ceiling. The movement interval allows prices up to $1,046.40, but the eligibility ceiling is $1,010. The combined numeric interval is therefore [$902.40, $1,010].

Step 3: Screen the menu. The $900 candidate is below $902.40 and fails the movement rule. The $930 candidate is within the combined interval and is eligible. The $1,000 candidate is also within the interval and is eligible. The $1,040 candidate is within the movement interval but exceeds the eligibility ceiling, so it is ineligible.

Step 4: Select the best surviving candidate. The feasible candidates are $930 and $1,000. Their contributions are $68 and $72, so the $1,000 candidate is preferred.

Check. The selected price changes by \((1,000-960)/960=40/960=4.1667\%\), which is within the movement limits, and it does not exceed $1,010. The $1,040 candidate’s movement alone would be acceptable, but that does not make it eligible. This example avoids treating one satisfied guardrail as a substitute for satisfying all of them.

Worked Example: Conflicting Limits and an Empty Feasible Set

A hypothetical policy has a current premium of $700. Its renewal may increase by no more than 10%, and a product-specific minimum-premium rule requires at least $780. The available menu is $750, $780, and $805. All three candidates pass the policy’s non-price eligibility checks. Their modeled expected contributions are $55, $64, and $70, respectively. Determine whether a recommendation is feasible.

Step 1: Convert the increase cap. The maximum allowed premium under the movement rule is \(700(1+0.10)=\$770\). There is no stated movement lower bound, so the movement rule does not create a higher minimum in this example.

Step 2: Combine with the premium floor. The floor requires \(P_j\geq \$780\), while the movement cap requires \(P_j\leq \$770\). A premium cannot satisfy both inequalities, because \(780>770\). The numeric feasible interval is empty.

Step 3: Confirm against the menu. The $750 candidate is below the required $780 floor. The $780 candidate meets the floor but exceeds the $770 movement cap. The $805 candidate also exceeds the cap and meets the floor. Therefore none of the menu entries is feasible.

Conclusion and check. No recommendation can be issued under the stated rules, despite the menu having an apparently attractive $805 candidate with contribution $70. The infeasibility follows from the conflicting bounds, not from a failure to search the menu. The correct action is to report the conflict for authorized resolution, not silently relax a guardrail or label the highest-contribution candidate feasible.

Common Mistakes and Implementation Checks

Movement rules are easy to misapply when a percentage, a dollar amount, and a menu entry appear together. In the first worked example, an incorrect calculation might treat a 12% maximum increase as a $12 increase and set the cap at \(800+12=\$812\). That is a percentage-versus-dollar error. The correct increase is \(0.12(800)=\$96\), giving a cap of \(800+96=\$896\).

Other errors arise from the order of operations or from treating eligibility as an optimization preference rather than a hard condition. The following checks help make the result auditable:

  • Use the stated base. Calculate percentage movement from the specified current premium, not from the candidate premium or from an unrelated portfolio average.
  • Keep signed changes straight. A -5% minimum change means a lower bound of \(0.95P_0\); it does not mean the new premium is 5% of \(P_0\).
  • Apply every rule. A candidate that satisfies the movement interval can still fail an eligibility test, premium floor, or ceiling.
  • Do not optimize over rejected candidates. First form the feasible set; then apply the customer-level objective to its members.
  • Make infeasibility explicit. If the bounds conflict or every menu item is excluded, state that there is no feasible candidate under the stated rules.

These checks make a quote-level recommendation traceable: each candidate’s status follows from a defined comparison basis, numeric bounds, and eligibility conditions. They also distinguish a genuine lack of feasible prices from a merely lower objective value.

Check Your Understanding

For each hypothetical renewal, apply all stated customer-level rules before choosing a candidate. The answer includes the feasible choice or states when none exists.

  1. A policy’s current premium is $1,250. Its movement range is -4% to +6%. The menu is $1,200, $1,240, $1,300, and $1,330, with expected contributions of $90, $95, $108, and $111. What is the feasible set and best feasible price? (Answer: Bounds are $1,200 to $1,325; feasible prices are $1,200, $1,240, and $1,300; select $1,300 with contribution $108.)
  2. A policy has a current premium of $840. Its dollar movement limits are a decrease of at most $35 and an increase of at most $50. An eligibility ceiling also limits the quote to $875. The menu is $800, $820, $870, and $890, with contributions of $40, $52, $61, and $70. Which candidate is best feasible? (Answer: Combined interval is $805 to $875; feasible prices are $820 and $870; select $870 with contribution $61.)
  3. A policy’s current premium is $1,500, and movement must be between -8% and +5%. The menu is $1,400, $1,500, and $1,550, with contributions of $73, $77, and $82. The $1,550 quote requires an endorsement that this policy is not eligible to receive; the other two pass eligibility. What is the best feasible price? (Answer: Movement bounds are $1,380 to $1,575; $1,400 and $1,500 are feasible; select $1,500 with contribution $77.)
  4. A policy has a current premium of $900, a maximum permitted increase of 10%, and a required minimum premium of $995. All candidates pass other eligibility rules. The menu is $980, $995, and $1,010. Is a recommendation feasible? (Answer: No; the movement cap is $990, which conflicts with the $995 floor, so the feasible set is empty.)
  5. A renewal has a current premium of $640 and a movement range of -3% to +7%. An eligibility rule also requires a premium no greater than $680. The menu is $620, $630, $660, and $690, with contributions of $40, $48, $56, and $63. Identify the feasible prices and select the best feasible candidate. (Answer: Movement bounds are $620.80 to $684.80; the $620 and $690 candidates fail the combined bounds, so $630 and $660 are feasible; select $660 with contribution $56.)
Clinical Biostats