The Power of Macroeconomic Predictions

As a professor of economics and a macroeconomic researcher, I am often asked to comment on the ‘economy’, national or global. Occasionally this means I am asked to explain the likely impact and advisability of different policy options, or to discuss long-run trends. For this I have at my disposal a body of economic theory buttressed by historical experience as my guide. Mostly, however, I am asked to provide predictions about the very short-term behaviour of some macroeconomic time-series, such as next quarter’s gross domestic product, the value of the FTSE 100 next week or the value of sterling next month. This is more problematic. It is, of course possible – though difficult – to make such forecasts, at least in a highly limited way. The real problem arises when I am asked to explain the underlying rationale behind my predictions. Given our present state of knowledge, we can forecast short-term phenomena but have no workable economic theory that can explain why things will go one way and not another, as we do for long-term developments. The crisis in confidence in macroeconomics that followed the global recession of 2007 directly results from the conflation in the public imagination, particularly in the business world, between models designed to explain the world and those designed to forecast it.

The confusion is hard to disentangle from the newest wave of nostalgia for the Keynesian consensus that emerged immediately following World War II and crashed so thoroughly during the era of 1970s stagflation. In those more optimistic days, Keynesian economics promised that government spending could act as a panacea that would banish the business cycle and ensure stable and uninterrupted growth forever. Painful tradeoffs were replaced by the paradox of thrift, and the idea that the more a nation’s citizens spent on consumption today, the more income would be available to enjoy tomorrow. And if that was not optimistic enough, the models – really no more than a few simple, though artfully chosen relationships – would not only help policy makers make all this operational, but could also forecast whatever short-term fluctuations might remain, predict the long-run growth of all the major sectors of the economy, and even explain in some seemingly intuitive way how all these different sectors interact.

It is a rather odd paradox that the more macroeconomists distance themselves from the last of these ideas and confront the uncomfortable fact that macroeconomic data cannot be understood without reference to the millions upon millions of complex human interactions they represent, the more we are perceived as arrogant simply because we have no choice but to rely on mathematics to cope with such complexity. No serious economic researcher today believes that we are close to devising macroeconomic models that can perform all the amazing tricks the post-war models purported to accomplish.

Put simply, there are two broad and mutually exclusive categories of macroeconomic models. The first type are based on carefully articulated economic theory, designed to carefully model the behaviour of individuals (workers, consumers and savers), firms and the government. Macroeconomists deploy sophisticated mathematics merely to construct and analyse the behaviour of models that in turn generate artificial time series with statistical properties we hope might match their real-life counterparts. Such models can also provide a useful platform to isolate the likely effects of changes to government policy. These models could not predict the last recession, or indeed the next one, as they were never constructed for this purpose at all.


So what about forecasting? Fortunately, the very same statistical properties of the macroeconomic variables that the aforementioned models are built to explain can be exploited to generate forecasts of these variables’ future values. This second category of models rely on theory to help choose which variables to include, but their structure is essentially atheoretical, as the functional relationships between the variables are highly simplified to facilitate estimation. The resulting models can in most instances outperform mere guesswork, but only as long the relationships estimated on the basis of previous history do not suddenly shift. Though this fact – that forecasting models are only as good as the relevance of the historical record upon which they rely – has been cited again and again by each new critic as a deep insight, it is a reality that economic forecasters know all too well. Much of empirical macroeconomics is devoted to finding trend breaks in historical data that result from fundamental shifts in the economy.

Just as the equations of quantum mechanics state that it is impossible to determine both the position and velocity of a particle, a model that abstracts from the messy and complex details of human interactions can still usefully predict aggregate behaviour – but not explain it. The more details these models try to incorporate, the more we sacrifice their usefulness as tools for forecasting. Call this the macroeconomic analogue to Heisenberg’s uncertainty principle.

For example, the latest macroeconomic forecasts produced by the International Monetary Fund (IMF) predict the world economy will grow at an annualised rate of 3.867% during the third quarter of 2010, and then by 4.419%, and the corresponding rates for advanced economies are 1.958% and 1.993%. What underlying processes or set of policies generate this pattern are unknown. The tools used by the IMF to predict next quarter’s output growth are statistical black boxes. Only those who do not understand how these forecasts are produced would be foolish enough to hazard a guess.

Using Macroeconomic Models

None of this implies that theoretical models are merely academic exercises, or can be used only as tools for understanding policy. On the contrary, macroeconomic models have great predictive power – but only for predicting and explaining long-run change. For example:

The normal rules of diminishing returns apply to China today just as much as they applied to the Soviet Union in the 1950s, Japan in the 1980s and the tiger economies of south east Asia in the 1990s. China will, in the not too distant future, face slower growth, and the slowdown may be far more abrupt and disruptive than business people in the West, in thrall to the vision of an ever-expanding market, may expect.

The sovereign debt crisis has only just begun, and Western policy makers have yet to address the true scale of the fiscal imbalances facing their aging societies. The relative generosity of the pay-as-you-go public pensions systems now in place is unsustainable. Even if policy makers can muster the political will to impose extremely onerous taxes on younger workers and savers, such actions will either fail to solve the problem or cause economic stagnation. Tax rates in western Europe are already high, and there is increasing evidence that some taxes are generating the maximum amount of revenue they can. Other taxes could be raised, and new ones imposed to generate greater revenue, but only by greatly disincentivising work and investment.

The temporary bailout of Greece devised by the European Union (EU) and the IMF this spring merely postpones the inevitable restructuring of its debt. This is well understood in the markets and is reflected in the high borrowing costs Greece is still paying. Nonetheless, despite the inherent deficiencies of the European Monetary Union (EMU), in particular its inability to impose even the modest fiscal discipline envisioned in the Maastricht treaty, the euro as a currency is here to stay for the foreseeable future. Despite the scare-mongering by the media, the banks and EU officials, there is simply no direct relationship between the value of the euro or the future of the single currency and the default on sovereign debt by one or more of the countries on Europe’s periphery, provided the European Central Bank (ECB) itself does not lose its nerve. Put simply, the obligations of the Greek or Portuguese government are obligations of the Greek or Portuguese government and no else’s. They do not belong to the ECB or its member states – assuming, of course, that (in an effort to protect these countries or lower their borrowing costs) the reassurances made by those who set policy for the EU implicitly make them so. US states have defaulted on their obligations in the past, California may do so soon, yet the dollar is still with us. No one thinks that if a county council in the UK finds itself unable to meet its immediate obligations, the Bank of England (BoE) will close its doors. The euro is no different – unless the European Commission (EC) persists in its self-defeating and highly risky strategy of asserting that it is.

The unsustainability of Western public finances (particularly in the US) and the economic future of emerging economies (particularly China’s) are not unrelated. The longer the US continues to borrow from abroad, the greater will be its incentive to monetise the accumulated debt. China can maintain its mercantilist policies as long as it is willing to finance Western standards of living in exchange for dollars or US treasuries that will buy less and less on world markets. By maintaining its undervalued fixed exchange rate, China will soon get to experience the joys of importing US inflation. China’s economy cannot sustain its present growth rates, nor have its policy makers abolished the business cycle. Just as Western policy makers may stimulate the economy today in exchange for worse problems tomorrow, the longer the Chinese delay revaluing their currency, the more difficult and sudden may be the crisis when their economy begins to slow. No forecasting model can predict the precise moment this reckoning will arrive, but we know from our theoretical models that it is on its way – and from historical experience that such crises do not necessarily emerge gradually or with any advance warning.


Related reading