Reinsurance Market Microstructure Don Mango Guy Carpenter 2 Capital Market Microstructure Major Components* Price formation and discovery: how latent investor demands are translated into realized prices and volumes Market structure and design: relation between price formation and trading protocols Information and disclosure: transparency, ability of market participants to observe information about the trading process *”Market Microstructure: A Survey,” Ananth Madhavan, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=218180 3 Continuous Double Auction Standard mechanism for price Buy limit orders are BIDS formation in most modern financial Sell limit orders are ASKS or OFFERS markets At any given time, there exists Two types of orders: – Market orders – requests to buy – BEST (lowest) ASK price and or sell a given number of shares – BEST (highest) BID price immediately at best available The difference is called the BID-ASK price (impatient traders) SPREAD – Limit orders – worst allowable price to transact with a time limit; Each Bid or Ask has the following not always immediately properties: a price, volume, and time transacted, so stored in a queue limit known as an order book Midprice = (BID + ASK)/2 4 Continuous Double Auction Figure 1 Theoretical Order Book Maximum depth = all available shares (stock) or notional outstanding (bonds) Quoting costs, herding effects limit the realistic range to be within certain bounds of Mid-Price – I.e., not feasible to produce infinite quotes for all possible prices BID to BUY Realistic Range Smooth Curve = Continuity Best Bid Mid-Price Best Ask ORDER SIZE AXIS Changes over time as quotes expire or are removed, or orders are filled HIGHER PRICE One such order book exists for each security LOWER PRICE Theoretical Order Book – Continuous = no price gaps – Deep = ability to satisfy any market order without price impact ASK to SELL Order Size = Depth PRICE AXIS 5 Focus on Realistic Range Figure 2 Realistic Order Book Transaction occurs when a Sell Order can be matched to a Buy Order BID to BUY Excess Supply Transaction LOWER PRICE Actual Order Book – Discrete not continuous = composed of individual quotes – Each quote represents the willingness of an individual market participant (agent) to buy or sell – Minimum Price increments = ticks – Order book can be sparse (have gaps) – Market makers are supposed to fill out gaps in order book, but this can be costly if they have to keep position net HIGHER PRICE ASK to SELL Excess Demand 6 Liquidity Crisis = Sell Off Figure 3 Liquidity Crisis Sell Off Not enough Buyers anywhere near the Mid-Price Sellers have two choices: 1. Be patient = hold their Asking price constant, wait for market to stabilize and liquidity to return (temporary market failure) 2. Lower their Asking price to the level necessary to find a Buyer Each lowering demonstrates impatience, creates incentives for Buyers to put new Orders even lower This is the mechanics of a price drop!! BID to BUY Demand Dried Up Excess Supply Mid-Price LOWER PRICE HIGHER PRICE ASK to SELL Large Gap in Order Book 7 Price Movement in a Continuous Double Auction “What really causes large price changes?” Farmer et al*, 2004 High density of limit orders per price (“full order book”) results in high liquidity for market orders implies small movement in the best price when a market order is placed Price movement is not uniquely defined, but midprice is often used Midprice can move due to arrival of: – Market order bigger (in volume) than the opposite best quote widens the spread by increasing Best Ask if it is a buy order, or decreasing Best Bid if it is a sell order – Limit order inside the spread – Cancellation of a limit order * www.santafe.edu/~baes/jdf/papers/fluctFinal.pdf 8 Price Movement in a Continuous Double Auction (cont’d) Liquidity = measure of market depth and continuity – Depth = amount of shares available – Continuity = orders close together, not spaced far apart Low liquidity can lead to large price movements when filling orders Depth of order book is a representation of individual investor appetite for positions 9 Demonstration of Price Movement Figure 4 Liquidity and Price Movement Example Filling a Market Sell Order for 600 Shares Under Low Liquidity and High Liquidity Bid # 1 2 3 4 5 Best Bid Price Low Liquidity Order Book Pre-Transaction Shares to Fill Bid Price Bid Shares Market Order 35 100 100 34 200 200 33 300 300 32 400 31 500 35 Bid # 1 2 3 4 5 Best Bid Price High Liquidity Order Book Pre-Transaction Shares to Fill Bid Price Bid Shares Market Order 35 600 600 34 600 33 600 32 600 31 600 35 Post-Transaction Bid # Bid Price 4 5 Best Bid Price 32 31 32 Post-Transaction Bid # Bid Price 2 3 4 5 Best Bid Price 34 33 32 31 34 10 Reinsurance Market Auction (RMA) Structure Not continuous but timed (effective date, renewal cycle) Synchronized blind auction (no way to see other Asks or Bid) There is an order book of Asks maintained by the Broker Three phases: (I) Price Exploration and Quote Development, (II) Asking Price Development, and (III) Firm Order Terms Two types of orders: – Bid = what a cedant thinks they should pay for the reinsurance – Quotes (Asks) = what a reinsurer offers to sell the reinsurance Each Bid or Ask has the following properties: a price, volume, and time limit Type of agent determines type of order: – E.g., Reinsurer does not Bid, only one Bid (from cedant itself) 11 RMA Phase I Price Exploration and Quote Development Figure 5 RMA Phase I Reinsurer 1 Submission Reinsurer 2 BROKER CEDANT Proprietary Portfolio Info Quotes (Asks) Reinsurer 3 … 12 RMA Phase II Asking Price Development 13 RMA Phase III Firm Order Terms Figure 6 RMA Phase III Firm Order Terms Reinsurer 1 BROKER Reinsurer 2 Reinsurer 3 Price and Strategy Evaluation CEDANT … Reinsurer n Desired Share 14 Arbitrage Opportunities in the RMA? Identification of a possible arbitrage? – Involves private contract between cedant and the reinsurers – Final value of this contract is private, so traded derivatives are unavailable. – No short-selling Can the RMA punish a reinsurer whose asking price is wildly divergent from the consensus range of quotes? Over-Priced Re might be asking more than other markets because: – Higher technical price due to a higher indicated layer loss cost, higher internal expense requirements, or a higher profit load; – Higer strategic differential due to a desire to nudge the final price upward or indicate weak interest. The RMA results for Over-Priced Re: a low (or no) share being offered. That’s the extent of the market punishment. 15 Price Movement in Reinsurance Auction Price moves due to changes in More difficult to define price Asking Prices: movement than in capital markets – Technical Price changes: Far fewer sequential data points innovations in loss cost estimates; (annual) increased profit margins (e.g., post Sept 11) Dissimilar products crosssectionally and over time – Strategic differentials – Product lines Blind auction, signals or anticipation of other actions – Cedants – Liquidity = market depth – Opaque differences in features Enough signed lines at a given – Different underlying portfolios price to fill out the program – Brokers estimate comparable pricing 16 Price Movement in Reinsurance Auction (cont’d) Could have some degree of consistency on approach to Technical Price derivation But there are many valid reasons why that would and perhaps should differ among competitors If Strategy differentials were zero everywhere, market quotes would at least reflect legitimate cost differences (where cost includes desired profit margin) and could be called “high information content” Informational Reductions: – Modification of technical price (esp. loss cost) to make market price appear more appealing – Strategy differentials: invisible changes in price that may or may not represent any “information,” merely positioning or other incentives 17 Reinsurance Market Liquidity Crisis 2006 U.S. Property Catastrophe Reinsurance “Perfect storm” of influences led the U.S. catastrophe reinsurance market to what can only be called a liquidity crisis U.S. insurers were unable to purchase reinsurance in the desired quantity at anything resembling the expiring prices Systemic crisis, striking across the board The RMA mechanics that led to this crisis were: – Catastrophe model changes, – Changes to rating agency capital formulas, and – Loss of retrocessional capacity. 18 Catastrophe Model Changes Reinsurers and brokers use catastrophe models for layer loss cost, program pricing and structuring. Insurers base their catastrophe reinsurance purchases on: – Key PMLs like 1-in-100 year and 1-in-250 year occurrence loss; – Prior year reinsurance purchasing, often defined in terms of program attachment and exhaustion return periods; and – Peer purchasing decisions, again in terms of return periods. Discipline around cat modeling is so ingrained in this market that variations in that variations in quotes (asking prices) among reinsurers is low Variations in quotes are due to internal expense loads, profit loads, and strategy differentials. 2006 RMS introduced version 6.0 of US Hurricane, leading to dramatic increases in PMLs and layer loss costs 19 Rating Agency Changes Fall 2005, A.M. Best changes BCAR formula. – Previous BCAR subtracted after-tax impact of one net catastrophe PML (one-in-100 wind event or one-in-250 earthquake event). – In mid-2005, A.M. Best introduced a stress test to monitor the impact of a second catastrophe event on the BCAR for all insurers. – Reinsurers responded by reducing limits in high catastrophe zones, as well as attempting to move exposures to retrocessionaires, sidecars or catastrophe bonds. Similarly, on March 21, 2006, Standard & Poor’s revised its criteria to include an exposure-based catastrophe capital charge for insurers, similar to the capital charges for reinsurers. 20 Results 1. PMLs increased 2. Required purchasing increased 3. Layer loss cost estimates increased 4. Available supply decreased –increased rating agency stringency and the loss of retrocessional capacity 5. Price for that reduced supply increased – due to the substantial deficits from 2004 and 2005, the owners of reinsurers targeted higher returns, which translated to higher profit margins underlying the quotes. Liquidity Crisis Many large U.S. insurers, with exposure across the country, were unable to place their desired programs. Could not buy the desired amount of limit even if they raised their bids. Liquidity breakdown was not a price issue, but a capacity issue. The supply of additional reinsurance capacity (cat bonds, sidecars) could not grow quickly enough. 21 22 Reinsurance Market Microstructure Don Mango Guy Carpenter