Economic Forecasting: Issues and Challenges

Economists have come under repeated criticism in recent years. In an era of refined economic theories supported by advanced models and tools, economists continue to face challenges in forecasting changes in the economy and forewarning crises reliably. The 2008 financial crisis, post-recession recovery and the economic impacts on the UK of the Brexit vote are three recent, high profile examples of misjudged forecasting. These events have shaken the confidence in modern economic theories and models.

It is important to understand the purpose of economic forecasting before evaluating its recent failure. Economists use economic theories and models to understand global systems and issues. An economic theory is an abstract representation of a complex system which captures the underlying key features whereas a model is considered to be an applied version or empirical representation used to drive a hypothesis that can be tested. Theoretical models offer predictions based on qualitative analysis and empirical models help to verify the qualitative predictions of theoretical models using mathematical calculations. The empirical models rely on input – independent variables (e.g. fiscal policy such as income tax rates) and output – dependent variables (e.g. inflation). The model coefficient determines the relationship between input and output variables – how input variable affects the output. Historical data is then used to predict the most suited coefficient for the relevant economic system. Most economists use these models for economic forecasting.

There are many reasons for economic forecasting errors. An article published in the guardian asked how, nearly a decade after 2008 crisis, economists managed to forecast incorrectly (and with confidence) the short term economic impacts of a vote in favour of Brexit on the UK economy. The article suggested that in both cases the economists were ‘guilty of groupthink.’ It goes on to say that the quality of predictions suffers if the predictions are politically biased or aimed to support a particular school of thought or interest. Andrew Haldane – the chief economist of Bank of England blamed the Brexit forecasting errors on economic models and their limitations including their inability to cope with “irrational behaviour”. Even the most sophisticated models do not work well in the event of shock. Haldane stated explained that models are only as good as the information they process- fed poor quality data, they will make poor quality predictions. Broadly speaking the data used for estimating the coefficient is based on current conditions. If a model is challenged with an entirely new situation for which it is unprepared then the forecasting tends to suffer. Therefore, economists should clearly provide explanations of the conditions under which certain models can be used.

Some critics went as far as to blame economics education. Emphasis given to more mathematical based models was criticised and lack of knowledge on philosophy, politics, theology and history is now seen as a weakness in current economics education.

In addition, the models’ assumptions have been criticised. In the case of Brexit, the Treasury used a three-staged approach. First, it used gravity models to estimate the impact of different trade relationships. Gravity models took geographical closeness and historic links between countries into account rather than the lowest tariffs. This could be seen as favouring negotiation towards European single market rather than working to remove all barriers to imports. Second, the model used relied on academic data – the efficiency of the UK economy from different levels of trade and foreign direct investment, to estimate the consequence of productivity. A recent Working Paper 483 from University of Cambridge pointed out that Brexit is a unique event. No other country has left the EU in the past so there is no experience or data to draw from, thus no better way for gathering realistic data for predicting the value. Thirdly, it used the productivity number calculated in the second stage in a global economic model to estimate the long-term differences in national income and prosperity. It is worth noting that the model assumed that policy makers and governments are not going to make adjustments or has not fully integrated the trade policies.

In conclusion, the main reasons for forecasting failures can be broadly categorised into limitations faced by models, poor quality data, flawed assumptions and inability to think outside mathematical models. The models are simplified frameworks, they lose the details, thus focus on the problem at an abstract level incomparable to the real world. Therefore, knowledge outside mathematics becomes beneficial in order to understand systems and issues. This can be used to arrive at valid and realistic assumptions which will help incorporate and provide a more sophisticated approach to economic forecasting.

Leave a Reply

Your email address will not be published. Required fields are marked *