Much of the world of actuarial science, particularly in insurance pricing, is an exercise in predicting the future. The costs we are trying to price for are not those of today, but 18 months from now. And while we know that these costs and market conditions are constantly changing, most insurers adjust their prices once, maybe twice, a year. It is widely accepted that making rate changes more often would be beneficial, but a deep dive into the underlying dynamics reveals a surprising truth: reducing the delay between your pricing analysis and the effective date of new rates can be a more impactful strategy.
This post explores the fascinating interplay of frequency, lag, and market volatility, drawing insights from an actuarial model designed to quantify the benefits of different pricing strategies.
The Temporal Challenge in Insurance Pricing
The core challenge in insurance pricing is temporal. It’s the difference between where you are and where you need to be. Pricing is like a quarterback throwing a pass: you don't throw to where the receiver is, you throw to where they are going to be.
In personal lines insurance, this prediction is fraught with difficulty due to several interconnected lags:
- Data Lag: When actuaries sit down to work, they rarely use the most recent data. There’s a period of time - three months is common and used in this exercise - to let claims get fully reported and begin to develop before the experience period is deemed ready for analysis.
- Analysis Lag: There is also the time it takes to analyze the data, build the models, and finalize the pricing decision. This can be many months, but for this exercise we are assuming straight-forward rate adequacy reviews completed in one month. At the end of the month there is an agreed decision on what the new prices should be.
- Implementation Lag: Then there is the period between the pricing decision and the effective date when the new rates are actually on the street. We set that here at five months, but it can be much longer depending on regulatory and IT realities.
- Earning Lag: Finally, a one-year policy takes a full year to earn its premium. This means a person who accepts the new rate on the last day it is offered is still locked in for another year. In this paper we are assuming one-year policy terms.
When all these factors combine, the average loss cost you are truly targeting for your new rate can be an astonishing 21 months or more after your experience period ends. This is what makes the prediction so hard, especially in a volatile market. We may revise rates once a year, but the future is a constantly moving target.
The Baseline Scenario: Normal, Plodding Pricing
To quantify the effects of different strategies, a baseline pricing strategy was established:
- Rate Change Frequency: Once per year.
- Implementation Lag: A nine-month gap between the end of the experience period and the effective date of the new rates – 3 months of development, 1 month of analysis, and 5 months for implementation.
- Inflation Environment: A stable 2% annual inflation rate (loss cost creep).
This model assumes actuaries can perform without error tasks like developing claims and trending historical data. The focus is purely on the structural constraints imposed by rate change frequency and implementation lag. What if the pricing strategy reviewed rates more often and/or shortened the time it takes to get new rates on the street?
Testing Volatility: The Inflation Spike
The effectiveness of pricing strategies was tested by introducing a significant, but temporary, spike in the inflation rate - a jump from 2% to 5% for one year, before returning to the 2% baseline.
The results for our baseline pricing strategy illustrated the danger of a slow, constrained pricing process:
- The Loss: As soon as the inflation rate spikes the insurer begins losing money, and on a calendar month basis, the underpricing peaks at roughly 2% of pure premium.
- Delayed Reaction: The actuaries only start to notice the change in inflation during their next annual rate review. Even then the change in historical loss costs hasn’t fully reached their experience period data.
- Overreaction: Just as the historical data is fully showing the effects on loss cost of higher inflation, the inflation rate has returned to normal. Again there is a lag in recognizing this, and the actuaries project another large rate increase for the next year, causing an overreaction. Now the insurer is charging too much, making them vulnerable to competitors who can react more quickly.
The analysis shows that in this volatile environment, the mispricing – measured as the deviation from the actual required rate – can linger for an incredible two and a half years after the market changes.
Quantifying Improvement: Frequency vs. Lag
To compare the effectiveness of different strategies, the cumulative effect of mispricing over time was quantified by measuring the area under the mispricing curve for the different pricing strategies. The improvements discussed are the reduction in this metric as compared to the baseline pricing strategy. An improvement means that either the size of the mispricing, or the length of time it was experienced, was reduced. Or, in the best scenarios, both.
Note that the exact timing of market changes and rate reviews can affect the percent improvement seen. The numbers shown below can change notably with different timings, but the relative relationships and the conclusions discussed – such as that the improvements are significant and some strategies outperform others – hold essentially constant.
Strategy 1: More Frequent Rate Changes
The first strategy tested was increasing the rate change frequency from once per year to twice per year, while keeping the five-month implementation lag constant.
- Effect: This allows the actuary to assess the state of play more often and react more quickly. The size of the mispricing doesn’t change much, but it works its way through the book of business more quickly.
- Impact: The change resulted in a 7% reduction in the mispricing area, mostly because the peak pricing error shifted forward in time.
Conclusion: More frequent rate changes allow for a faster recognition of market changes, but they don’t address the inaccuracies inherent in predicting the future.
Strategy 2: Shorter Implementation Lag
The second strategy focused on reducing the time between the pricing decision and the effective date, from five months to two, while keeping the rate change frequency at once per year.
- Effect: By predicting a nearer-term future, the pricing is inherently more accurate, and the size of the distortion is less. In addition, there is a slight speed-up in the time at which the peak pricing error is seen, though not as much as in the twice-yearly rate change case.
- Impact: This strategy resulted in a 25% reduction in the mispricing area, due to both the more accurate pricing and the increased speed of it working through the book.
Conclusion: The effect of reducing implementation times for new rates is, in this model, significantly larger than the effect of increasing the frequency of rate changes.
Strategy 3: The Synergistic Approach
The true benefit comes from tackling both constraints simultaneously – twice-yearly rate changes with faster implementation times.
- Effect: Smaller distortions in pricing than for either single approach, that work their way through the book of business more quickly than for either single approach.
- Impact: The combined strategy showed a 38% reduction in the mispricing area, greater than the sum of their individual effects.
Conclusion: In terms of the size of the mispricing and the length of time the mispricing is experienced, making both improvements results in better outcomes than for either change alone. There is a real synergy in accomplishing both.
The Ultimate Vision: Continuous, Fast Pricing
Due to the use of historical data, and the length of the policy term, some distortion in pricing caused by market changes is unavoidable. To understand what can be achieved from more responsive pricing, we pushed the model to the limits of its framework.
- Continuous revisions: which in this model means monthly changes in the rates.
- No implementation lag: we kept the three-month lag for loss reporting and development, and also allowed a month for analysis. But once a pricing decision was made, this scenario assumes that the rates are available on the next day.
This aggressive strategy resulted in a 61% reduction in the mispricing area. The distortion in the rates was reduced to less than 1%, and that mispricing worked through the book over a year more quickly.
A Call to Action for Actuaries and Regulators
The model’s results affirm the benefit of both more frequent rate changes and faster implementation times. It also distinguishes differences in the size and effects of each change, allowing insurers to make more informed decisions for where to focus resources and energy.
- Focus on speed: If you have to choose, the priority should be finding ways to implement new rates faster. Insurers should invest in tools and processes that dramatically reduce the time it takes to get an approved rate factor into the market.
- Increase frequency where possible: The ability to change rates more often is usually limited by internal and external resource constraints. More efficient tools and processes can reduce the effort required by internal pricing teams. More proactive partnership between insurers and regulators may facilitate more dynamic and frequent reviews.
To make changes like these, insurers will need to invest, but the potential for benefit is large. The underlying risks, market competitors, and economic and regulatory environments all change more or less constantly. This is true in highly regulated markets as well as ones that allow for daily changes in rates.
Wherever you are currently with your pricing strategy, this data shows that the closer we get to the ideal of quick implementation and frequent review, the more resilient and profitable our pricing will be. For more information, please visit the Guidewire PricingCenter website.