The fifth anniversary of the start of the most severe financial crisis since the Great Depression has us reflecting on how little we were able to predict from macroeconomic models – the very models we use to capture business cycle dynamics. While being careful not to completely overturn the mainstream body of economics, the very fact that they could not predict the most important economic phenomenon in nearly a century requires us to question the way these models perform.
The focus of most Dynamic Stochastic General Equilibrium (DSGE) models used by central banks is to keep inflation at its target value. Before the crisis, most countries were experiencing high rates of inflation. Consequently, central banks responded aggressively by raising interest rates to keep pace. Thus, central banks literally restrained the economic activities in these countries by following the inflation targeting models designed to deal with high inflation. Complications arise when we consider that this assumption is extensively focused on the distortions in price levels and its association with low-to—medium inflation rates. However, in reality the extent of misfortune created by the financial crisis is not comparable to the potential costs of high inflation.
There are several reasons that have led macroeconomists to this level of disarray. Decades ago, economists started noticing that there is no clear-cut connection between macroeconomics and microeconomics, leading to increasing attempts to connect these two fields. We saw the rise of macroeconomic models based on microeconomic foundations incorporating assumptions of perfect competition and perfect information in the market. This is indeed the concept behind DSGE models.
These assumptions, and more importantly the “representative agent” as an assumption leading to the impossibility of internal lending, indeed turned out to be critically wrong. The problem with the models based on the representative agent assumption is that there is literally no financial sector in these models. In other words, either the whole sector defaults or no one defaults. Furthermore, in these models all the market participants are as credible as the government. So a bank’s IOU’s can be exchanged as money. There is no money or central banks involved in these models.
Not surprisingly this debt-inflation problem indeed proved to be one of the strong explanatory variables in predicting the crisis.
The nature of the financial system links the crisis to the debate about the role of central bank and finance companies and the inadequacy of their regulatory systems. While DSGE models would address cyclical perspectives, every day decisions of financial institutions involve liquidity management practices, namely buying and selling securities and operating in the short-term money market. And the crises came as a product of individual banks’ liquidity shocks. Even though the connection of banks’ balance sheet shocks to monetary policy was widely recognized by economists ahead of the crises, regulatory aspects of this conclusion that could have possibly prevented these shocks where disregarded.
How this has changed since that time? And have predictive models attempted to incorporate such aspects?
If banks can still largely fund their assets at their pleasure with borrowed money, other systemic crises will follow. In addition, dysfunctional regulatory frameworks will add fuel to the fire.
In light of such considerations, and with the 2008 crises having proved the amplifying effects of the financial sector on business-cycle fluctuations, the G20 has taken a new approach to risk and regulation in the financial sector by endorsing the new Basel III capital and liquidity requirements. The concept of liquidity buffers was introduced, and in order to avoid the repeat of a “liquidity fortress” scenario the liquidity coverage ratio was established. Asymmetric information, in the form of financial frictions and macro prudential policy concepts, has been incorporated into new general equilibrium framework models.
For researchers, the critical task going forward is identifying the unrealistic assumptions behind macroeconomic models and to correct for, and possibly include, the financial sector in all DSGE models. And we need a more focused microeconomic data analysis of related macroeconomic questions.
In the spirit of the current financial crisis, we have learned a lot about macroeconomic modeling, which makes us optimistic about the future of modeling in the macroeconomic research agenda. However, the learning process is still at an early stage if we look at it through the perspective of what will remembered as a historic financial moment. Therefore this conversation needs be continued.
