FORECASTING is a tricky game. Subject to the whims and fancies of unforeseen events, the future can be illusive. Yet, when the big property agents cannot agree in their assessment of Hong Kong's residential sector, questions are raised. This has proved the case with the current series of reports which estimate what happened to property prices in the luxury sector during the third quarter. Bears had their suspicions confirmed by First Pacific Davies, who reckoned that prices had dipped by 5.5 per cent, while bulls could take comfort from Brooke Hillier Parker's pronouncement of an average 14 per cent increase. Jones Lang Wootton said there had been a modest rise of 2.24 per cent for medium-to large-sized units, while Land Power said the correction averaged seven per cent across the whole of the residential market. These figures all relate to the top-end sector, even if the weighting each firm gives to its sub-divisions varies. Bearing in mind that the agents' raison d'etre is not only to execute transactions but also to provide insight into both a property's current and future value, some worrying implications arise. It follows that if the territory's top-dog valuers cannot agree on the general level of prices in the recent past, variations can be expected in current valuations, not to mention forecasts. For the agents, a clubby silence is being kept on the subject. Other than a few blandishments to the effect that different firms are looking at different sectors of the market, few are prepared publicly to criticise their competitors. In explaining the differences, a series of overlapping theories emerge. The most popular is that the last quarter saw a fall-off in the level of transactions, meaning that the sample size on which agents founded their figures was much smaller. As any statistician will confirm, a smaller sample implies a greater degree of volatility in the estimates extracted from it. It is true that, during the last quarter, transactions slipped, due to the effects of the banks' tight mortgage criteria and a tailing off in mainland buyers, but this is not the whole story. Other possibilities include the large numbers of renovated properties which have been coming on to the market, raising the question of how to price the increase in value resulting from redevelopment. Despite these factors, it is more likely that the abnormal market conditions in the third quarter simply magnified the variations inherent to firms using different compiling techniques. Merely collecting data on deals done over the period and aggregating them out will not give a true market representation. In more liquid times, the error resulting from this kind of methodology will be small. Yet when activity is low, oddities such as the sharp increase in the book value of Swire's Robinson Place units, due mainly to low pre-sale prices, can easily distort a price index. Such distortions, which arise from a lack of liquidity, can be seen by comparing a time-series of asking prices with the actual prices paid for properties. When times are good and buyers and sellers plentiful, the two sets of figures tend to tally. When bad, and characterised by a lack of liquidity, there are wide variations.