My last post on this blog concerned Okun's law. In the graph below, I report on the level of growth that is needed to maintain stable unemployment, calculated using 20 year windows from UK data from 1972-2010 and based on the model referred to in my earlier post. Over most of this period growth has needed to be just a whisker under 2 per cent per year to ensure that unemployment does not rise. In the last two periods illustrated on the graph, this has fallen dramatically - to just a little over 1 per cent.

It is not clear why, and it is certainly not clear that this drop represents a permanent shift.

In 1962, Arthur Okun wrote his famous article about the relationship between the change in the unemployment rate and the growth rate of real output. This included the statistical estimation of an equation that relates changes in the unemployment rate to the growth rate of GDP. With some rearrangement of the terms in this equation, it is straightforward to evaluate the extent to which the economy needs to grow in order to ensure that unemployment does not rise. This relationship has come to be known as Okun's law - and it indicates that, in Western economies, somewhat over 2 per cent growth per annum is needed just to keep the unemployment rate steady. In the absence of such growth, advances in productivity will mean that output can be produced with fewer labour resources than were previously required.

Recent data to emerge from the UK suggest that this law is being repealed. Output has been flat in recent months, but unemployment (far from rising, as Okun's law would predict) has been falling. In some respects this is easy to explain: workers are accepting wage settlements that are below inflation, so that the brunt of adjustment appears in the form of falling real wages rather than rising unemployment. Several factors may account for this: communication about the extent of the global readjustments required may have given workers a hefty reality dose; the union legislation of the 1980s, now really being tested for the first time, may have proven itself to be effective.

Increased union militancy challenges this last possible explanation. And however strong the reality dose might be, it is more likely than not that at some point resistance to real wage cuts will strengthen. Okun's law has been one of the more durable statistical regularities in economics over the years, and growth remains the most promising way of avoiding large increases in unemployment.

I have, from time to time, posted on this blog forecasts of the UK economy that come from a neural network model using monthly index of production data. The latest forecast - based on data up to April of this year - appears in the red line of the graph below. The blue line are historic data, while the forecast looks ahead 24 months.

This forecaster has, for many months now, consistently predicted a dip in economic activity towards the end of this summer. Like any economic forecasts - indeed more so than most, given its simple nature - it comes with a 'health warning'. While some economic indicators - most recently the drop in the unemployment rate - provide welcome news, there is still some way to go before we can be confident that growth has been restored.

The International Monetary Fund has reported on its mission to the UK. The report argues that the 'curent settings of fiscal and monetary policy remain appropriate in the central scenario', but that 'risks and uncertainty around this central scenario are significant'. In particular, if growth remains weak and if unemployment remains high, the automatic stabilisers that are built into fiscal policy - including lower tax take and higher spending on benefits - should be allowed to operate freely.

As early as May of last year, there were signs that the government intended to be flexible in the way in which it implemented the fiscal retrenchment. The Chancellor of the Exchequer has made much of the idea that there is no plan B, But the rhetoric is now changing - a plan B seems to be embedded within a plan A that is more permissive than the government's own rhetoric has often implied. For the sake of those who bear the brunt of this ever-so-slow recovery, let's hope so.

In the beginning, there were real business cycle models of the kind pioneered by Kydland and Prescott. These attempted to provide a rationale for the existence of smooth business cycles by examining how rational, utility maximising economic agents respond to random shocks by redistributing their activity over time. Eventually these models developed into what we now call dynamic stochastic general equilibrium (DSGE) models. These models initially adopted the market clearing assumptions made in the original real business cycle models. More recently, models have become more general and have introduced the notion of staggered price adjustments. The latter models are often referred to as new Keynesian variants, and the work of Michael Woodford is typical of this endeavour.

All of the models referred to above are designed to build up a story of macroeconomic behaviour using a strong set of microeconomic foundations. They are based on individual economic agents that are acting rationally. By drilling right down to the preferences of agents, the macroeconomic behaviour of the models is supposed to finesse the problem raised by the Lucas critique - the notion that, since a change in policy might change the way in which people respond to policy, models based on these responses cannot be used as forecasting tools. (But since people's preferences don't change in response to policy - only their behaviour does - models built on what we know about people's preferences are deemed to be robust.) In consequence, DSGE models have attracted much interest amongst central bankers.

But DSGE models are difficult to work with. Studying individual decision making in a dynamic environment - where attention has to be given to the way in which people form, and respond to, expectations about the future as well as how they respond to the past - involves a lot of complicated analysis. Simplifications have to be made. This being so, early DSGE models have focused on output and prices. While results from some of the models have been encouraging, many economists - myself included - have been underwhelmed by the general approach. This is because the models have failed to consider a world in which some people are unemployed. In other words, they have failed to consider anything that looks, even remotely, like the real world.

All that is set to change. Jordi Galí, one of the pioneers of new Keynesian DSGE models, has recently published work that accommodates the existence of unemployment. This has been built upon in further work done in collaboration with Frank Smets and Rafael Wouters. It's still early days, but the promise of DSGE modelling looks as though it may well be realised.



Galí, J. (2011). THE RETURN OF THE WAGE PHILLIPS CURVE Journal of the European Economic Association DOI: 10.1111/j.1542-4774.2011.01023.x
Kydland, F., & Prescott, E. (1982). Time to Build and Aggregate Fluctuations Econometrica, 50 (6) DOI: 10.2307/1913386
Lucas Jr, R. (1976). Econometric policy evaluation: A critique Carnegie-Rochester Conference Series on Public Policy, 1, 19-46 DOI: 10.1016/S0167-2231(76)80003-6
Woodford, M., & WALSH, C. (2005). INTEREST AND PRICES: FOUNDATIONS OF A THEORY OF MONETARY POLICY Macroeconomic Dynamics, 9 (03) DOI: 10.1017/S1365100505040253