- Time and Date: 10:00-16:55, Friday December 19, 2008 (Open: 9:45 a.m.)
- Venue: RIETI's seminar room (1121,11th Floor, METI ANNEX)
1-3-1 Kasumigaseki, Chiyoda-ku, Tokyo
- Language: English (no interpretation available)
Session 2: "Nominal Rigidities, News-Driven Business Cycles, and Monetary Policy" (with NUTAHARA Kengo)
KOBAYASHI Keiichiro (Senior Fellow, RIETI)
The news-driven business cycle (NDBC) is defined by positive co-movements in output, consumption, investment and labor when positive news about the future arrives. This model is studied because it may hold an explanation for certain boom and bust episodes, including the Japanese real estate bubble of the late 1980s and the U.S. Internet bubble of the late 1990s. The traditional real business cycle (RBC) model is inadequate in explaining these co-movements.
Given a function for the evolution of technology, news is defined as a kind of technology shock. The technology follows an autoregressive process with two components; the shock observed at time t and the shock observed at time t - p . The second component is the news shock.
The standard RBC is made up of one utility function and a production function. News shocks on positive future productivity shocks do not change the technology parameters or capital stocks. However, as agents want to increase consumption, labor input must be reduced. This is a mechanism that the standard RBC cannot generate as can the NDBC. In order to generate the NDBC, the intratemporal condition must be violated.
There are several strategies to change the intratemporal condition. NDBCs can be generated without market failures. Authors attempting this approach have changed production technology or preference functions to create a complicated form of intra-temporal conditions. For example, Beaudry and Portier use multi-sector production technology and complementarity between consumption and investment goods. NDBCs can also be generated with market failures. In this case, market failures are adopted to generate pliable labor wedges between marginal rates of substitution and marginal products of labor. Den Haan and Kaltenbrunner use matching friction in the labor market.
This paper proposes nominal standard rigidities as a new mechanism for NDBCs using sticky prices and adjustment costs of investment. This model can generate NDBCs from both technology growth and level, and procyclical movements of Tobin's q. It also generates recessions if the news regarding technology growth shocks turns out to be false. Decreases in the markup cause co-movements of consumption, labor, investment and output.
The model is a standard New-Keynesian sticky price model. Households have a utility function distinct from labor, and a function for adjustment costs of investment. The final-goods sector is competitive, while the intermediate-goods sector is monopolistically competitive. Two functions are used for production technology. Also, Calvo-pricing with price indexation is used. The monetary authority adopts the forward-looking Taylor rule.
To investigate the performance of the model, a Christiano, Ilut, Motto and Rostagno (CIMR)-type experiment was used. Before t = 0, the economy is at a steady state. At t = 0, news arrives of a 1% increase in technology growth. At t = 4 or one year later, agents know that the news turns out to be false. A recession occurs when the news turns out to be false. The parameters of the model are the same as CIMR, although there are a few exceptions.
From this experiment, certain things can be learned about growth. Co-movement was observed between consumption, output, Tobin's q , rental rates, investment, labor and wage rates. When positive news arrives, future consumption increases and future labor supply decreases due to the future wealth effect. However, in order for the future labor supply to increase, the future wage rate should increase and the future markup must decrease. As a result of this, the competitiveness of the intermediate-good sector will increase. There will be an increase in future price, and due to this there will also be an increase in the optimal price.
However, agents cannot change their optimal prices due to nominal rigidities, and the markup today will decrease as a result of sticky prices. This causes aggregate demand, output and labor input to increase. Finally, investment and consumption increase due to a loosening of household budgets. When the news turns out to be false, some of the aforementioned factors are reversed. Markups increase as a result of sticky prices, but aggregate demand, output and labor input, as well as investment and consumption all decrease, similar to boom conditions. Because of nominal rigidities, these changes lead to recession.
The second case is a news shock to the technology level. Co-movement was seen between consumption, output, Tobin's q , investment, labor and wage rate. The two key differences with the growth experiment are the delay of responses and the absence of recessions when the news turns out to be false. The key factor is the adjustment cost of investment, which causes increases in investment and consumption today. Also, an increase in aggregate demand causes an increase in labor input and a decrease in markups.
Finally, the results were compared with those of CIMR, who also employ sticky prices. As for the differences between the two models, the model used in this study removed habit persistence and sticky wages from what was reported in the CIMR paper, and modified the adjustment cost of investment. CIMR found that the role of habit persistence and flow adjustment was very important to the NDBC; however, they show that Tobin's q does not move procyclically in their model. In order to produce procyclical movement of Tobin's q , CIMR needs to add sticky prices, sticky wages, and the inflation targeting rule for monetary policy. They do not check whether nominal rigidities alone can generate the NDBC. This study found that nominal rigidities alone can generate NDBC. Furthermore, it was found that friction between nominal rigidities and habits violate the intratemporal condition, and that sticky wages are also a mechanism of NDBCs.
To summarize, nominal rigidities are a new mechanism for generating NDBCs. They use the standard New-Keynesian model along with sticky prices and adjustment costs of investment to produce an NDBC. The key factor is the countercyclical movement of markups. This model generates NDBCs from news about both technology growth and level, and pro-cyclical movements of Tobin's q . It also generates recessions if the news regarding positive growth turns out to be false.
Question and Answer Session
Q. Please explain the lack of empirical evidence in the study. While it could be stated that Beaudry and Portier provide this evidence, that is not entirely true.
Please explain the markup dynamics further. In a sticky price model, the technology shock would cause the markup to go up.
If news is not private information, why is there not a provision for a change in monetary policy with a news shock? Monetary policy should react to the news if it is publicly available.
I agree that there is not much empirical evidence in this study. I took a very theoretical approach, but there are several projects currently underway that are identifying news shocks in Japanese and U.S. data. Hopefully, evidence can be found that news shocks are important in business cycle fluctuations or other boom-bust episodes in the U.S. or Japan. That being said, news-driven business cycles may not be very important empirically. This paper is only a theoretical exercise to show that this simple mechanism can produce co-movements.
As for the movements of markups, in response to positive technology level shocks the New-Keynesian model produces procyclical markups. In response to positive growth rate shocks, as in this study, markups decrease.
This experiment is quite singular, in that news is assumed to be false. The news shock is completely killed by the news shock at time t - p . To provide a simple illustration on the performance of the model, a very simplified experiment was used. The monetary policy rule also may be incomplete.
Q. If the model is correct, a news shock cannot be a dominant source of business cycles. How do you explain the low volatility of real wages in regard to news shocks?
In response to a 1% shock, the wage rate saw a 2% response, which is too volatile. I agree that the model needs substantial improvements to explain the real episodes in history.
Q. It seems that an analysis of the Japanese economy would be best carried out with a comparison of models with and without habit persistence. Please explain how the study dealt with habit persistence
Some experiments were done with and without habit persistence. Today, only one or two results of those experiments were presented. Even if we introduce the habit persistence into our model, our results do not change. Also, an assessment needs to be made to determine whether or not habits are more important than sticky prices.
Q. Have you tried shutting down the investment sector to see what kind of dynamics you get out of the growth news shocks when you just have labor as a factor of production?
When the investment sector is shut down, the dynamics will be reduced to the simplest version of the New-Keynesian model. It is apparent that consumption, labor, and output co-move in such a model. Shutting down the investment sector would lead to trivial results.
Q. The technology growth news shocks lead to very short recessions in this study. Can this study explain longer recessions? In subsequent tests there are no recessions. Is this due to adjustment costs of investment or is it due to a short persistency parameter?
This is a very simple model and as such is not able to explain the decades-long stagnation of Japan of the 1990s. More and different factors need to be added to explain longer recessions. This model shows that in response to news turning out to be false, there will be a very sharp, short-term recession. If Japan's lost decade is to be explained, completely different factors need to be added to this model. This model can explain the bubble period of the 1980s in Japan.
Regarding the second question, level shocks are too small compared to growth rate shocks. A one-time shock in the growth rate changes the level of technology permanently, whereas a level shock changes the technology level only temporarily.