ANZ Research published its insights and prediction of Sydney House Prices for the rest of 2021. With forecasted gains of 23 per cent in the 2021 calendar year, this is up from their previous prediction of between 15 to 20 per cent in March 2021. If you go back even further than that, this prediction is up from their November prediction of 9% for 2021.
I predict that there will be future predictions that are wrong
ANZ Research expects “housing price gains of around 9 percent across the capital cities. An early COVID-19 vaccine rollout and the resulting lift to sentiment could drive larger price gains than currently anticipated. That said, ANZ Research thinks regulators would be quick to step in with macro-prudential measures if the market looked to be overheating.”
End to end, the prediction was completely wrong. The act of ‘revising’ a forecast is admitting while not admitting that you were wrong. I’m not blaming anyone for being wrong, no one on this earth has a crystal ball and can accurately predict the future.
So why do we continue to indulge in predictions and forecasts? The same reason people visit fortune tellers, people want certainty in uncertain times. A forecast can help provide people make decisions when the future is uncertain even when these forecasts are wrong by providing confidence in decision making. However providing forecasts can be misleading, where they entice decisions based on events that haven’t occurred.
The problem lies at its foundation. The researchers employ the use of models. Modelling attempts to build an abstract representation (a model) of a real world situation and uses it to make predictions based on the likely movement of variables. The movement of these variables and their interaction are the ‘laws’ of economics but they aren’t really laws. In science, laws of nature are unshakeable like the law of entropy or gravity. If I drop a ball at height it will fall to the ground. Laws in the economy interact more like guidelines in a man-made conception. This is why economic crisis occur unforeseen. From this shaky foundation, models are built to predict the future.
Models only work when data is input into it. The subjective bias of the researcher whether they are aware of it or not influences the outcome of the model based on their understanding of the interaction between variables. These assumptions can lead down a path where findings can be manipulated. For example if the model of future house price growth returned a forecast of 50% growth. This figure seems unrealistic even to an everyday person, but do you question data inputs or their interaction in the model? Is the model incorrect? if so, how and why?
I doubt any researcher would put their name to a forecast that seems to the human eye a bit over the top. So what would the researcher do? Manipulate the model or the input variables to reveal a more believable number?
Researchers understand that the models are not accurate. There is a direct competition between the real world and the elegant world of mathematics where everything makes sense and every variable is accounted for. The real world does not care about how perfect the mathematics is. Models are trying to map through quantitative analysis how humans behave and interact with changes in economic conditions and changes in the property market. Which is not an accurate measure of the reality of how humans behave.
The nature of the property market generally shows long periods of growth with short periods of no growth or negative growth. The falls in growth usually are more dramatic than the increase in growth. This very rule has been broken as we have seen exponential growth in the last 6 months.
As models use inputs to make predictions. The model are based on variables staying the same for their model to be true. As we all know change occurs all the time and change drives reality. What we have seen in the past is that sometimes the risks that have the smallest probability of occurring. Have the greatest impact because no one ever saw it coming.
As a researcher, it makes more sense to predict that the market would continue along. this is because more than likely that is what will happen. It is better to predict that everything will continue as it did today and be right most of the time. Than to predict doom and gloom all of the time and be catastrophically correct once.
The best thing that we can take from this, is not to read forecasts as anything more than a continuation of the present situation without any changes which as we know is never the case.
Forecasts can be used to fit into a particular narrative that the news source wishes to convey as people can misinterpret forecasts as hard facts. Perhaps we need to come to terms with ourselves. Knowing that we are living in uncertain times and perhaps a forecast providing us with reassurance is the adult equivalent of a mother telling her son that ‘everything is going to be okay.’ Is not the best way to deal with risks and making decisions.