Skip to main content

Detecting Financial Imbalances: Monitoring Financial Imbalances through the Financial Activity Indexes (FAIXs)

Koji Nakamura, Yuichiro Ito (Bank of Japan)

Research LAB No.15-E-1, March 19, 2015

Keywords:
Financial imbalances; bubble; early warning indicator; financial crisis

JEL Classification:
E44, G01

Contact: kouji.nakamura@boj.or.jp (Koji Nakamura)

Summary

Financial imbalances such as asset price bubbles and excess credit expansions tend to lead to financial crises and associated abrupt credit crunches. To prevent such a disaster, it is important for national authorities to detect financial imbalances at an early stage. This note explains the "Financial Activity Indexes (FAIXs)," a set of indicators developed by Bank of Japan staff which are used to detect financial imbalances. The FAIXs consist of 14 macro indicators. Each indicator is judged to be "overheating" or "overcooling" by evaluating the extent to which the actual levels of the indicator exceed certain thresholds. Overall developments are summarized in a "heat map," a visual representation of financial imbalances. If more "reds" are observed in the "heat map," then financial imbalances are judged to have reached an excessive level.

1. Introduction

Financial imbalances such as asset price bubbles and excess credit expansions tend to lead to financial crises and abrupt credit crunches. We have observed many episodes of accumulated financial imbalances and subsequent financial crises in both advanced and emerging economies, including the global financial crisis triggered by the bankruptcy of Lehman Brothers, the collapse of Japan's bubble economy in the 1990s, and the Asian crisis in 1997.

Based on past experience of financial crises, authorities around the globe have recognized that it is important to detect financial imbalances at an early stage and to take preemptive measures to forestall potential financial crises. To this end, national authorities and international organizations have committed to developing and utilizing early warning indicators of financial crises to detect financial imbalances such as overheating in financial activities (Borio and Drehmann (2009)). This note explains the "Financial Activity Indexes (FAIXs)," which are used to detect financial imbalances in Japan. The details of the FAIX compilation methodology were described by Ito et al. (2014). The latest FAIX conditions and heat map are shown in the Financial System Report (FSR), a semi-annual report on Japan's financial system. The latest issue of the FSR was published in October 2014.

2. Compilation of Financial Activity Indexes

The FAIXs were compiled as follows.

First, we collected as many candidate indicators as possible based on early studies. We used 159 indicators as candidates for this analysis.

Second, we sorted these indicators into appropriate categories based on economic entities and activities. We set up six economic entities: the financial institution sector; the financial market sector; the private sector; the household sector; the corporate sector; and the real estate sector. We chose the real estate sector because the activities of this sector were closely related to land prices and credit movements, and because these activities were different from those of the household and corporate sectors. Regarding activity-based categorization, we established two categories: investment activity on the asset side and funding activity on the liability side. Furthermore, we established two asset price categories -- stock and land prices -- in addition to the above. As a result, we have 14 categories. We split the 159 indicators into these 14 categories.

Finally, we chose the most appropriate indicator from each category based on statistical evaluations. In this process, we decided on appropriate trend extraction methods and relevant thresholds, above which we judge each indicator to be showing signs of "overheating," among several methods and thresholds.

We evaluated the indicators based on two statistical criteria. The first criterion is whether each indicator detects overheating in financial activities during Japan's bubble period, which had a major impact on economic and financial activities in Japan. If the actual level of an indicator is above a certain threshold during the bubble period, the indicator appropriately detects overheating in financial activities. In this analysis, we assume on the basis of early studies that Japan's bubble period ran from 1987 to 1990.

The second criterion is whether statistical errors attendant to forecasting future events are minimized by using these indicators. The first type of statistical error is that an event occurred but no signals are provided by the indicator in advance. This is a risk of missing a crisis and is referred to as Type I error in statistical terminology. The second type of error is one in which no event occurred but signals are provided. This is a false signal risk and is known as Type II error in statistical terminology. When we set the thresholds at lower levels, signals tend to be provided more frequently and the risk of missing a crisis thus falls, but the risk of false signals increases. On the other hand, when we set the thresholds at higher levels, signals tend not to be provided and the risk of missing a crisis thus increases, but the risk of false signals lessens. As we saw, there is a trade-off between these two types of statistical errors. We calculated appropriate thresholds by balancing these two types of statistical errors.

We chose the best indicator from each category to minimize statistical errors with appropriate thresholds.

3. Selected indicators

We selected the 14 indicators as follows (Figure 1). Note that each indicator has a different trend and thresholds. This is the result of the statistical evaluations mentioned above.

Figure 1: Selected indicators

Figure 1: Selected indicators
Investment Activity (Asset side) Funding Activity (Liability side)
Financial Institutions DI of lending attitudes of financial institutions
<past averages, 1σ>
Growth rate of M2
<one-sided HP filter, 1σ>
Financial Markets Equity weighting in institutional investors' portfolios
<3-year moving averages, 1σ>
Stock purchases on margin to sales on margin ratio
<3-year moving averages, 1σ>
Private Sector Private investment to GDP ratio
<3-year moving averages, 1σ>
Total credit-to-GDP ratio
<one-sided HP filter, 1σ>
Household Household investment to disposable income ratio
<3-year moving averages, 1σ>
Household loans to GDP ratio
<3-year moving averages, 1.25σ>
Corporate Business fixed investment to GDP ratio
<one-sided HP filter, 1σ>
Corporate credit to GDP ratio
<3-year moving averages, 1σ>
Real Estate Real estate firm investment to GDP ratio
<one-sided HP filter, 1σ>
Ratio of real estate loans to GDP
<one-sided HP filter, 1σ>
Figure 1: Selected indicators
Stock Price Land Price
Asset Prices Stock Prices
<one-sided HP filter, 1.5σ>
Land Prices to GDP ratio
<3-year moving averages, 1σ>

Notes:

  1. The trend extraction methods and thresholds are shown in <>.
  2. We consider three trend extraction methods: past average; three-year moving average; and a one-sided HP filter. In a one-sided HP filter, the HP filter is applied to individual sets of data leading up to the beginning of each period and the most recently filtered value is plotted. We set the smoothing parameter of the HP filter at 400,000.
  3. σ is a unit that represents the root mean square of deviation between actual and trend values.

We now explain the indicators selected for each economic entity.

Financial institution sector

Investment activity on the asset side of the financial institution sector is measured by the DI (Diffusion Index) of lending attitudes among financial institutions. Lending attitudes among financial institutions become more aggressive and the DI moves toward accommodative conditions as economic and financial activities become buoyant. Funding activity on the liability side of the financial institution sector is measured by the M2 growth rate. The M2 growth rate rises as financial institutions become more active in their lending activities and credit creation expands.

Financial market sector

Investment activity on the asset side of the financial market sector is measured by the equity weighting in institutional investors' portfolios. The denominator of this indicator consists of not only equities, but also bonds. As stock markets soar, institutional investors increase their stock weightings in their portfolios and the indicator rises. Funding activity on the liability side of the financial market sector is measured by the stock purchases on margin to sales on margin ratio. As stock markets soar, investors become actively engaged in stock purchases using margin credit and the indicator increases.

Private (non-financial) sector

Investment activity on the asset side of the private sector is measured by the private investment to GDP ratio. Private investment includes capital investment, inventory investment, residential investment and durable goods consumption. Funding activity on the liability side of the private sector is measured by the total credit to GDP ratio. The total credit to GDP ratio is an important reference indicator when national authorities determine the level of the counter-cyclical capital buffer to be introduced under Basel III (see Drehmann et al. (2010)).

Household sector

Investment activity on the asset side of the household sector is measured by the household investment to disposable income ratio. Household investment includes residential investment and durable goods consumption. Funding activity on the liability side of the household sector is measured by the household loans to GDP ratio.

Corporate sector

Investment activity on the asset side of the corporate sector is measured by the business fixed investment to GDP ratio. Funding activity on the liability side of the corporate sector is measured by the corporate credit to GDP ratio. Corporate credit includes bank lending, corporate bonds and trade credit.

Real estate sector

Investment activity on the asset side of the real estate sector is measured by the real estate firm investment to GDP ratio. Real estate firm investment includes business fixed investment, land investment, and inventory investment. Funding activity on the liability side of the real estate sector is measured by the ratio of real estate loans to GDP.

Asset prices

We adopt two asset price measures: stock and land prices. Stock prices are represented by the TOPIX and land prices by the land prices to GDP ratio. Land prices are measured by the urban land price index of six large city areas for all uses.

4. Heat map

Overall developments in the selected indicators are summarized by a "heat map," a visual representation of financial imbalances (Figure 2). In the "heat map," areas shaded in red show that an indicator has risen above the upper threshold, areas shaded in blue show that an indicator has declined below the lower threshold, and areas shaded in green show everything in between. Areas shaded in white show periods without data. During the bubble period in Japan, every indicator is red, suggesting overheating, although the duration of the red period for each indicator varies. Around the mid-2000s, before the Lehman shock, only real estate investment shows red and suggests overheating. Therefore, the degree of financial imbalances is judged to be low. Recent indicators show no signs of overheating in financial intermediation as most indicators are green.

Figure 2: Heat map

  • Heat map(1980 to 2014). The details are shown in the main text.

We also look at detailed developments among several indicators. Stock and land prices move above the upper thresholds for a long time during the bubble period, as shown by the shaded areas in the following figures (Figures 3 and 4). Stock prices have recently moved just above the trend, but have remained within the thresholds. Land prices have been around their trend.

  • Figure 3
    Graph: Stock Prices. The details are shown in the main text.
  • Figure 4
    Graph: Land Prices to GDP ratio. The details are shown in the main text.

Similar to asset prices, the private investment to GDP ratio and the total credit to GDP ratio exceeded their upper thresholds for an extended period during the bubble period (Figures 5 and 6). Most recently, both indicators have moved just above their trends but have remained within their thresholds.

  • Figure 5
    Graph: Private Investment to GDP ratio.  The details are shown in the main text.
  • Figure 6
    Graph: Total credit-to-GDP ratio. The details are shown in the main text.

Regarding the financial sector, the DI of lending attitudes among financial institutions moves beyond the upper threshold before the bubble period (Figure 7). Even during the bubble period, this indicator continues to signal overheating. The M2 growth rate stays above the upper threshold for a long time during the bubble period (Figure 8).

  • Figure 7
    Graph: DI of lending attitudes of financial institutions. The details are shown in the main text.
  • Figure 8
    Graph: Growth rate of M2. The details are shown in the main text.

The following points need to be noted. First, the selected indicators were chosen based on the past episode of the bubble period. Therefore, it might be difficult to detect financial imbalances by using the existing FAIXs due to new financial activities and forms of intermediation such as more direct linkages with overseas economies. In order to capture potential risk factors related to newly developed financial activities, it is important to collect relevant data by using anecdotal information from financial institutions. Second, we need to pay attention to various interactions among financial activities. We concentrate on developments among individual indicators when using the FAIXs. However, based on past financial crises, it is important to understand various interactions such as the feedback loop between the real economy and the financial sector and transactions among financial institutions. To do this, it is important to use macro stress testing taking account of the feedback loop between the real economy and the financial sector. See Kitamura et al. (2014) for details of the Bank of Japan's macro stress testing program.

Reference

Notice

The views expressed herein are those of the authors and do not necessarily reflect those of the Bank of Japan.