Kobayashi-sensei's Economic Research Picks

Part 15: What Brings About Long-term Changes in Expectations? The mechanisms by which the financial crisis has led to secular stagnation

Faculty Fellow

Econo-kun is in his second year of the master's program at a private university, studying hard to become an economist.

Kobayashi Keiichiro's photo KOBAYASHI Keiichiro: This time I will outline the mechanisms by which a single financial crisis brings about persistent economic stagnation as explained by Kozlowski, J., L. Veldkamp, and V. Venkateswaran in their paper "The tail that wags the economy: Belief-driven business cycles and persistent stagnation" (2015).

Secular stagnation theory and "changes in expectations"

Just as the question of how to modify deflationary expectations is a major topic in discussions about financial policy in Japan, the detailed analysis of the mechanisms that form expectations has been an important research agenda recently in macroeconomics.

This paper provides a theoretical model that uses changes in expectations to explain changes in the U.S. economy since the financial crisis (2008-2009).

Since the financial crisis, U.S. gross domestic product (GDP) dropped 12% below the pre-crisis trend and has not shown any indication of returning to its previous level. Ordinarily, when a recession ends, GDP reverts to its original trend level. Consequently, it may be said that the financial crisis has brought about a long-term structural change in the U.S. economy. The authors' hypothesis is that "the experience of a tail event, the financial crisis, has changed expectations and altered performance of the U.S. economy over the long term."

Econo-kun's photo Econo-kun: Why is a one-time financial crisis able to alter expectations over the long term?

Kobayashi Keiichiro's photo Kobayashi: According to the standard rational expectations hypothesis, such change does not happen. The authors eased the rational expectations hypothesis just a little to make it more realistic.

Recessions and financial crises are thought to be triggered by exogenous shocks (in this paper, defined as a shock where there is a depreciation of capital stocks). Standard rational expectations hypothesis assumes that "people know ahead of time the probability distribution function of exogenous shocks." Based on this hypothesis, even if a financial crisis arises, it is a shock that occurs based on a known probability distribution function, so the expectations (shock probability distribution function) held by people after the crisis dissipates will be exactly the same as that before the crisis. Therefore, the structure of the post-crisis economy may be said to be the same as that of it pre-crisis. In this way, the standard rational expectations hypothesis states that a financial crisis does not bring about long-term stagnation of an economy.

Expectations revised by learning from experience

The authors' hypothesis is that "people do not know in advance the true probability distribution function of exogenous shocks." Although this loosens the rational expectations hypothesis, it is a very realistic hypothesis. The authors based their work on the hypothesis of learning which states that "people look at the time series of previous instances (samples) of exogenous shocks and estimate the probability distribution function of shocks using the same methods as econometricians do based on the time series." The method for estimating the probability distribution assumes the use of the kernel density estimation procedure, which is widely used in non-parametric econometrics.

This technique forms an image of the estimate which is created by plotting the frequency of numerous samples on a graph and connecting these points in a smooth curve resembling the pattern of a probability distribution function.

In a model where the probability distribution function is revised based on such learning, a permanent mark is left in the probability distribution function where such experience is presumed. This is because the sample of a one-time shock will be used permanently when estimating subsequent probability distribution functions. So, when a financial crisis occurs, any subsequent tail risk will be assessed larger perpetually.

Econo-kun's photo Econo-kun: Could you please explain that in layman's terms?

Kobayashi Keiichiro's photo Kobayashi: Up until 2007, people didn't have any experience with financial crises, thus they were unable to experience one and learn from it and had assumed that there would be zero tail risk. However, between 2008 and 2009, people experienced an enormous financial crisis and started to believe that financial crises do occur roughly once in 80 years. The change from the expectation that "financial crises will not occur at all" to "a financial crisis may occur (once in several decades)" represents a major difference. This contrast explains the long-term difference in the U.S. economy between the period until 2007 and the period since 2009.

The authors use numerical simulations to show quantitatively that this long-term change in expectations is able to explain the change in the U.S. economy (long-term deviation from pre-crisis trend). In other words, it can explain that a short one-time financial crisis has changed expectations over the long-term, thereby plunging the U.S. economy into secular stagnation.

Why has the credit spread returned?

There is a persuasive counterargument to the theory that the financial crisis is the cause of secular stagnation. If the financial crisis were the cause of secular stagnation, then the difference between the loan rate and the risk-free rate (the credit spread) would also have risen over the long-term since the financial crisis. However, the credit spread promptly fell to pre-crisis levels in the real U.S. economy. Accordingly, this idea says that the financial crisis is not the cause of secular stagnation.

Econo-kun's photo Econo-kun: How does this paper refute the idea that the financial crisis is not the cause of secular stagnation?

Kobayashi Keiichiro's photo Kobayashi: The authors conducted simulations using the model to show that the credit spread promptly returned to pre-crisis levels. Using their model, changes in long-term expectations have brought about persistent stagnation in consumption and investment on the one hand, while the credit spread quickly returned to pre-crisis levels. In other words, they showed that the financial crisis is the cause of secular stagnation and that the credit spread reverts to its previous low level at the same time.

The cause of the credit spread declining after the financial crisis is explained as follows.

Changes in expectations (people coming to expect that the probability of another financial crisis is higher than they previously anticipated) have led to companies and households reducing their consumption and investment, and decreasing their borrowings for such purposes. Because total borrowings have decreased, the probability of loan defaults has fallen. The result is that credit spreads offered on debt defaults have also fallen. In this way, the result of the financial crisis has been a stagnant economy, but credit spreads on loans have fallen to pre-crisis levels.

Explainable or unexplainable based on past experience

This paper argues that beliefs about tail risks have changed permanently based on past experience: the financial crisis. As a result, demand (consumption and investment) stagnates, which is observed by the fact that the level of GDP falls permanently by a certain amount. Although the GDP level falls, the GDP growth rate does not fall. This point is consistent with what has taken place in the U.S. economy since the collapse of Lehman Brothers.

Econo-kun's photo Econo-kun: Is this point also consistent if we use the example for the Japanese economy?

Kobayashi Keiichiro's photo Kobayashi: The mechanism addressed in this paper is not able to explain the fall in the GDP growth rate that has occurred since the collapse of Japan's bubble economy.

For Japan, a significant alternative issue is the decrease in the growth rate. It is important that Japan's problem cannot be explained using the theory put forth by Veldkamp and her colleagues that expectations have been influenced by past experiences. If Japan's issue were defined as "an ever-growing anxiety about the future," then this theory might be applied to explain the drop in the economic growth rate. Anxiety about the future refers to a tail risk caused by concerns about economic chaos due to a worsening fiscal situation. This tail risk increases in proportion to the increase in government debt. If tail risk increases annually, then consumption and investment will become increasingly inefficient each year and the economic growth rate will fall. In other words, this is a scenario where anxiety about the future exacerbates expectations and brings about economic stagnation.

If expectations are influenced by past experiences, then, when a major surprise alters expectations, the problem will resolve itself. That ought to have been the basic philosophy of the quantitative and qualitative monetary easing. If expectations have deteriorated not because of past experience but are due to anxiety about the future fiscal situation, then no matter how many times the Bank of Japan surprises markets with additional easing, expectations will never stop deteriorating.

It is important to determine which case applies to the reality.

>> Original text in Japanese

October 20, 2016

Article(s) by this author