KR20140133341A - Method and apparatus for predicting industry risk using industrial warning signs - Google Patents

Abstract

The present invention suggests a method of predicting an industrial risk using economic indexes according to the size of industrial groups for financial institutions to preemptively manage risks. The present invention stores a plurality of selected signal indexes in a database to explain the risk state of each mid-level industrial group for each of a plurality of mid-level industrial groups which is segmented from high-level industrial groups; calculates signal error values showing how accurately each signal index represents the actual risk state of each industrial group and the threshold value for the signal error values; determines a comprehensive risk index of each mid-level industrial group based on the signal error value of the signal index of each middle-level industrial group; and proposes a method of predicting risks of industrial groups for preemptive rick management.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for predicting an industrial crisis using a crisis alarm signal,

The present invention relates to a method for predicting industrial crises using economic indicators.

Financial institutions need to analyze the stability of a company before providing it to the enterprise. In order to analyze the stability of the enterprise, the stability of the industry to which the company belongs is taken into consideration and the industrial stability can be predicted by taking advantage of the possibility of the industrial crisis.

When a financial institution manages a credit, the delinquency rate or rate of delinquency is considered to be a kind of industry risk factor. In other words, as the delinquency rate increases, the financial institution will decide to reduce the size of the loan. However, there is a problem that it is difficult for financial institutions to reduce the size of their loans when the industrial crisis has already occurred.

Unlike general industry indicators, credit indicators in the financial period, such as delinquency rates, may change due to preemptive risk management and policies of financial institutions. In addition, there may be a difference in recognizing the industrial crisis. For example, the point at which financial institutions perceive industrial crises, such as delinquency rates and default rates, is usually late relative to external variables that precede industry crises.

Therefore, financial institutions continue to worry about what economic indicators can predict the industrial crisis to anticipate and respond to industrial crises.

According to the present invention, in order to predict the industrial crisis for the purpose of credit management, the financial institution estimates a moment exceeding the threshold value using the leading indicators according to the size of the industrial group, generates a signal indicating the industrial crisis, And proposes a method and apparatus for predicting industrial crisis.

The method of predicting the risk of an industrial group for preemptive risk management according to an embodiment of the present invention is a method for predicting risk of an industry by grouping a plurality of signal indicators selected to explain a crisis state for each of the sub- Storing; Calculating a signal error value indicating a degree of each signal indicator correctly indicating an actual crisis state of a corresponding industry group and a threshold value of the signal error value for each of the signal indicators; And determining the total crisis index of each of the multiple classification industry groups based on signal error values of the signal indicators and whether the signal error values exceed a corresponding threshold value for each of the secondary industry groups.

The step of calculating the threshold value of the signal error value and the signal error value according to an embodiment includes: determining a first signal indicator among the signal indicators and a first crisis indicator generated due to the first signal indicator; Determining the signal error value indicating a probability that the first crisis indicator corresponding to the first signal indicator incorrectly represents the actual crisis state of the industry during the window period from the time of occurrence of the first crisis indicator; And determining a threshold of the first signal indicator that minimizes the signal error value.

According to an exemplary embodiment, the step of determining the total crisis index of each of the multiple classification industry groups may include comparing at least one signal error value exceeding the corresponding threshold value with a corresponding threshold value individually determined for each signal error value, Summing the values; And determining a negative value of the summed signal error value as a value corresponding to the integrated crisis index of the middle classification industry group.

An industry group crisis prediction apparatus for preemptive risk management according to an embodiment of the present invention includes a database for storing various indicator data for predicting industrial crisis for the preemptive risk management; A signal indicator managing unit for periodically selecting a plurality of signal indicators for describing a crisis state for each of the plurality of classified industry groups and storing the signal indicators in the database; A signal error value indicating a degree of correctness of each signal indicator of an actual industry in each of the signal indicators; a signal error value management unit for calculating a threshold value of the signal error value and storing the signal in the database; And a middle classification total crisis index determining unit for determining the total crisis index of each of the multiple classification industry groups based on signal error values of the signal indicators and whether the signal error values exceed a corresponding threshold value for each of the secondary industry groups.

The present invention proposes a computer-readable recording medium on which a program for implementing an industrial cluster crisis prediction method according to an embodiment is recorded by a computer.

According to the industrial crisis prediction apparatus or the industrial crisis prediction method according to various embodiments of the present invention, various industries are classified into industrial groups according to the types of industries, and industrial crises are analyzed individually for each industry group. Thus, It can provide forecasts optimized for specific industries rather than index-based crisis forecasts.

In addition, suitable leading indicators may be pre-selected to determine the industrial crisis of a given industry group, and the leading indicators that can be observed before the actual industrial crisis occurs can be analyzed. If it is observed that a crisis signal is generated in the leading indicators, it is possible to anticipate the possibility before a real industrial crisis occurs. Therefore, preemptive risk management is possible for each industry group.

In addition, since the leading indicators suitable for judging the industrial crisis of a certain industry group are selected in advance from the candidate groups of the leading indicators with the leading indicators having the minimum signal error value and good crisis prediction power, the analysis time required for the leading indicators Can also be saved.

In addition, since leading indicators and crisis indicators to explain the industrial crisis are databaseed in advance and the method of determining the comprehensive crisis index is made logic, it is possible to predict a consistent industrial crisis.

FIG. 1 illustrates a block diagram of an industry risk prediction system for preemptive risk management in accordance with one embodiment.
FIG. 2 illustrates a flowchart of an industry military crisis prediction method for preemptive risk management according to an embodiment.
FIG. 3 illustrates the determination factors of an industry military crisis prediction apparatus according to an embodiment.
Figure 4 illustrates the large and small category industries according to one embodiment.
FIG. 5 illustrates a detailed flowchart of an industrial cluster crisis prediction method according to an embodiment.
6 illustrates a chaotic matrix representing a matching relationship between a signal indicator and a crisis indicator in accordance with an embodiment of the present invention.
7 shows a relationship between a signal index and a signal error value according to an embodiment.
FIG. 8 illustrates the leading and window periods of a signal indicator and a crisis indicator according to an embodiment.
FIG. 9 shows a graph of the total crisis index of the multiple classification industry group according to an embodiment.
FIG. 10 shows a graph of a total crisis index and a trend of a non-interest loan portion of a middle class industry according to an embodiment.
11 illustrates a candidate set of signal indicators according to one embodiment.

1 illustrates a block diagram of an industrial crisis prediction apparatus 100 for credit management according to an embodiment.

The industrial crisis prediction apparatus 100 according to an embodiment includes an index database 110, a signal index management unit 120, a signal error value management unit 130 and a middle classification crisis index determination unit 140.

The industrial group crisis prediction apparatus 100 according to an exemplary embodiment of the present invention includes an index database 110, a signal index management unit 120, a signal error value management unit 130, and a middle classification crisis index determination unit 140 And a processor (not shown). Alternatively, the index database 110, the signal index management unit 120, the signal error value management unit 130, and the subclass integrated crisis index determination unit 140 are operated by respective processors (not shown) May operate as a whole, the industrial crisis prediction apparatus 100 may operate as a whole. The signal index management unit 120, the signal error value management unit 130, and the multi-category comprehensive crisis (not shown) according to the control of the external processor (not shown) of the industrial crisis prediction apparatus 100 according to one embodiment. The exponent determiner 140 may be controlled.

Various index data may be stored in the index database 110 according to an embodiment to determine the crisis state of an industry group. In addition, the index database 110, the signal index management unit 120, the signal error value management unit 130, and the subclass total crisis index determination unit 140 according to the embodiment may use various index data And a data storage unit (not shown) for storing input / output data generated in the processing. The industrial crisis prediction apparatus 100 may include a memory control unit (not shown) for controlling the data input / output of the data storage unit (not shown) and the index database 110.

The industrial crisis prediction apparatus 100 according to one embodiment can communicate with a central server, a branch terminal, a branch terminal, etc. of a bank via a data communication network. A network according to an exemplary embodiment includes an open Internet, a wired Internet network including a closed intranet, a wireless Internet network interworking with a mobile communication network, a TCP (Transmission Control Protocol) / IP (Internet Protocol) And possible communication means such as a computer network capable of various data communication including data communication based on the same protocol. For example, the industrial group crisis prediction apparatus 100 can perform data communication based on an IP socket method or a web socket method.

Industry (industry) is an economic activity organization that produces goods or services. Depending on the type of goods produced, the type of industry can be classified. There are many industries in the industry. Each of these industries has industry specific characteristics.

In the case of the Republic of Korea, the industry is classified into 20 categories. The industry classification of the Ministry of Commerce and Industry is divided into major, middle, and minor categories. We generally refer to the sub-category, which is classified by type of industry.

Each major industry group can be subdivided into a number of subdivided industry groups. For example, the industry group is divided into 26 major industry groups, and the major industry group can be classified into 98 subgroup industry groups. Among these industries, the industry group predicted the crisis for preemptive risk management by the industrial crisis prediction apparatus 100 according to one embodiment are the middle class industry groups.

The industrial crisis considered in the industrial crisis prediction apparatus 100 according to the embodiment may be an economic situation in which the financial health of the lending transaction company in the financial period is deteriorated to increase the delinquency rate. In addition, the economic situation in which the asset value of a company belonging to a certain industry declines or the financial structure deteriorates due to a change in government policy can be interpreted as an industrial crisis.

The industrial crisis can be different from the actual industry depending on the concentration and quality distribution of the sample by industry. In addition, unlike industrial indicators, credit indicators in the financial period, such as the delinquency rate, may fluctuate due to preemptive risk management and policies of financial institutions.

Accordingly, in order to predict a crisis index indicating an industrial crisis, the industrial crisis prediction apparatus 100 according to an embodiment classifies industries into large-category industries according to characteristics of each industry, and large- After subdividing into industrial groups, we can analyze the industrial crisis. The industrial group crisis prediction apparatus 100 according to an embodiment can predict the industrial crisis from the sub-category industries.

In addition, industrial crises such as delinquency rates and default rates can usually occur later than external variables that predate the industry crisis. In order to overcome such a phenomenon, the industrial group crisis prediction apparatus 100 according to an embodiment analyzes the preceding indicators that are representative of the actual industrial crisis and predicts the crisis indicators using the preceding indicators.

The operation of the industrial crisis prediction apparatus 100 according to an embodiment will be described below with reference to FIG. 2 illustrates a flow diagram of an industry risk prediction method 115 for preemptive risk management in accordance with one embodiment.

In step 125, the signal indicator management unit 120 according to one embodiment can select a plurality of signal indicators for explaining the crisis state for each of the multiple classification industry groups by a plurality of the multiple classification industry groups. The signal indicator management unit 120 according to an exemplary embodiment may store items of signal indicators selected for the classifier industry in the database.

In step 135, the signal error value management unit 130 according to the embodiment can calculate a signal error value indicating a probability that each signal indicator correctly represents the actual crisis state of the corresponding industry group, for each signal indicator. Also, the signal error value management unit 130 may calculate a threshold value of the signal error value by collecting the signal error value of the past predetermined period for each signal indicator.

In step 135, the signal error value management unit 130 according to the embodiment can determine a threshold value of a signal error value and a signal error value for each signal indicator.

More specifically, the signal error value management unit 130 can determine the first signal index generated by the first signal index and the first signal index among the signal indexes.

The signal error value management unit 130 can analyze the probability that the first crisis indicator corresponding to the first signal indicator during the window period from the generation of the first crisis indicator does not correctly represent the actual crisis state of the industry group . This probability value can be determined as a signal error value. The signal error value management unit 130 according to an exemplary embodiment may determine a threshold value of a first signal indicator that minimizes a signal error value.

The signal error value management unit 130 according to an exemplary embodiment of the present invention detects a signal error value from a first signal indicator at a time point preceding a first periodic signal index It is possible to analyze whether or not the first crisis indicator at the time point following the period indicates an actual crisis state.

The signal error value management unit 130 according to an exemplary embodiment of the present invention detects a first error in which a real crisis state does not occur but a crisis occurs in the first crisis indicator during a window period, The signal error value can be determined using the ratio of the second error that does not occur.

In step 145, the multi-classification integrated crisis index determining unit 140 according to one embodiment determines whether or not the signal error values of the signal indicators and the signal error values exceed the corresponding threshold values for each of the sub- The crisis index can be determined.

The multi-classification integrated crisis index determining unit 140 according to an embodiment may compare a signal error value with a corresponding threshold value individually determined for each signal error value and add at least one signal error value exceeding the corresponding threshold value . The multi-classification integrated crisis index determination unit 140 may determine a complement value (1-signal error value) to 1 of the summed signal error value as a value corresponding to the integrated crisis index of the middle classification industry group.

Credit scale data, representing the size of the loans in each subdivided industry group, can also be stored in the database. The industrial crisis prediction apparatus 100 according to an embodiment of the present invention may be configured such that the industrial crisis prediction apparatus 100 according to an embodiment of the present invention includes an industrial crisis prediction unit 100, Can be monitored for a crisis in the industry when the total crisis index of the industry exceeds the threshold.

1 and 2, operations for predicting the crisis of each of the plurality of sub-category industries have been described by the industrial crisis prediction apparatus 100. FIG.

3 to 11, various indicators to be considered in the process of determining the total crisis index of the sub-category industry group will be described in detail.

FIG. 3 illustrates the determinants of an industrial crisis prediction apparatus 100 according to an embodiment.

The industrial crisis prediction apparatus 100 according to an embodiment can analyze the crisis indicator 12 indicating that the credit situation of the characteristic industry is in a crisis state in order to predict future crises in a specific industry. The industrial crisis prediction apparatus 100 is an indicator that indicates a crisis indicator 12 indicating that the credit situation of a specific industry is in a crisis state and is an indicator indicating the crisis indicator 12 using industrial data 14 related to the overall specific industry. Can be predicted. There may be a large number of industrial data 14 as a leading indicator for one crisis index 12. The industrial group crisis prediction apparatus 100 can analyze the risk data of the crisis indicator 12 by generating a crisis composite index 18 by analyzing the industrial data 14 as a leading indicator.

For example, the industrial crisis prediction device 100 can use industrial data 14 such as Gross Domestic Product (GDP) or trade balance as an independent variable. From the independent variables, we can derive a crisis composite index or a composite crisis index (18) using independent variables to estimate the movement of a crisis index (12), such as the future rate of delinquency, which is a dependent variable.

The industrial crisis prediction apparatus 100 according to an embodiment calculates a total crisis index 18 in order to empirically or learningly control the degree of applying the independent variable and the dependent variable in determining the total crisis index 18, The user input model parameter 16 may be introduced.

As described above, the crisis indicator indicating the crisis state of a specific industry group in the industrial military crisis prediction apparatus 100 according to the embodiment is a dependent variable, and the signal indicator, which is a leading indicator for predicting the crisis indicator, Same as. It should be noted here that the risk indicators and dependent variables will be mixed in the description flow, but the crisis index and the dependent variable refer to the same subject. Likewise, signal indicators, leading indicators and independent variables in the flow of the description will be mixed, but all refer to the same subject.

The industrial crisis prediction apparatus 100 according to an embodiment can be classified into a crisis index (dependent variable) representing a crisis state and a leading indicator (independent variables) of a crisis index depending on the characteristics of a specific industry, Indicators and leading indicators for them can be set individually. Accordingly, in order for the industrial crisis prediction apparatus 100 according to an embodiment to determine the crisis index of the industrial group, it is preferable that the industrial group is classified.

Industrial crisis prediction apparatus 100 according to an embodiment can divide industries into two or more middle class industries. The industrial crisis prediction apparatus 100 according to an exemplary embodiment can determine a total crisis index for each of the multiple classification industry groups.

FIG. 4 illustrates a large classification industry group and a middle classification industry group according to an embodiment.

The industry group to be a crisis prediction target of the industrial military crisis prediction apparatus 100 according to one embodiment may include a shipbuilding industry 22 and a shipping industry 24. The shipbuilding industry (22) and the shipping industry (24) are the subdivided industry groups in which the shipbuilding industry of shipbuilding industry is subdivided when the shipbuilding industry is one of the major industry groups.

The industrial crisis prediction apparatus 100 according to an embodiment can determine the total crisis index of the shipbuilding industry 22 and the shipping industry 24 of the middle class industry.

Specifically, it is as follows.

The signal index management unit 120 of the industrial crisis prediction apparatus 100 according to an embodiment of the present invention may be configured to determine the subclass comprehensive crisis index of the shipbuilding business 22 by using a dependent variable for the shipbuilding industry and an independent variable And the signal error value management unit 130 may determine the signal error values of the corresponding independent variables 21.

In addition, the signal index management unit 120 selects the independent variables 23 that can precede the dependent variable and the dependent variable for the shipping business to determine the integrated classification crisis index of the shipping industry 24, The management unit 130 may determine the signal error values of the respective independent variables 23. [

The middle classification total crisis index determining unit 25 for the shipbuilding business 22 and the middle classification total crisis index determining unit 27 for the shipping business 24 are each configured to comply with the subclass comprehensive crisis index of the industrial crisis prediction apparatus 100 according to the embodiment And may correspond to the determination unit 140.

The middle classification crisis index determining unit 25 according to an embodiment can determine the middle classification crisis index of the shipbuilding industry 22 using the dependent variable and the independent variables 21 for the shipbuilding industry 22. In addition, the multi-category comprehensive crisis index determining unit 27 according to an embodiment can determine the multi-category aggregate crisis index of the shipping industry 24 using the dependent variable and the independent variables 23 for the shipping industry 24.

Therefore, the subclass aggregate crisis index determining unit 25 can individually calculate the aggregate crisis index for each subclass industry group.

Hereinafter, detailed operations of the signal index management unit 120, the signal error value management unit 130, and the middle classification crisis index determination unit 140 will be described in detail.

FIG. 5 illustrates a detailed flowchart of an industrial cluster crisis prediction method according to an embodiment.

In step 31, the signal index management unit 120 according to one embodiment removes irregular elements of the index data including the signal index as the independent variable, the crisis index as the dependent variable, or removes the abnormal value . Irregular elements or abnormal values cause errors in the model estimation and may be distorted considerably by the results of the interpolation, which is the next step, so it is desirable to remove them through the filtering step.

Through the filtering step, the graph of the signal indicator and the crisis indicator can be flattened. The flatness of each indicator graph can remove many irregular elements, while the original data can be further distorted. Therefore, it is desirable that the indicators be filtered to the appropriate line.

In step 32, the signal index management unit 120 according to an embodiment may perform an interpolation step to process the signal index and the risk index, from which irregular elements have been removed, through the filtering step into data of a predetermined period. Through the interpolation step, the empty data can be estimated from the data currently held. Therefore, all of the data at the time when data is missing or not disclosed according to a predetermined period can be estimated. Yearly data or quarterly data may be changed to monthly data.

In step 33, the signal index management unit 120 according to one embodiment can process different signal indicators into the same monthly data through an interpolation step. Different signaling indicators need to be converted into the same monthly data. For example, in the case of daily data, it is necessary to decide which data in the month will be used as data representative of the month. In addition, when only annual data or quarterly data are acquired, it is also necessary to decide how to fill each month data in the year or each month data in the quarter.

In step 34, the signal index management part 120 according to one embodiment can confirm, through the stability test step, whether the indicators collected up to step 33 are variables worth analyzing statistically. Statistically, a 'stable variable' refers to a variable with a pattern from which future variables can be estimated from past variables. Therefore, since the unstable variable does not need to be analyzed by the industrial crisis prediction apparatus 100 according to the embodiment, it is necessary to transform the unstable variable into a stable variable.

The signal index management unit 120 according to an exemplary embodiment can replace unstable variables with stable variables through an augmented dicky fuller (ADF) test. For example, the difference between successive variables can be a stable variable, even if it is not stable, as shown in Table 1 below.

Date An unstable variable Stable variable 2011/02 10 2011/03 15 5 (= 15-10) ... 13 -2 2011/08 20 7 2011/09 15 -5

If the independent variable is replaced by a stable variable through the stability test step according to Table 1, the first data of the independent variable and the dependent variable may be removed together since the first data of the independent variable may be eliminated and the total number of data may be reduced .

In step 35, the signal error value management unit 130 according to the embodiment can determine a signal error value for each signal indicator through an independent variable NTS calculation step. The signal error value NTS (noise to signal) is a value indicating the possibility that the risk identification capability of the corresponding independent variable is inaccurate. Therefore, the signal error value can be used as a value to confirm how well the signal indicator predicts the crisis indicator.

For example, the greater the frequency at which a signal indicator does not anticipate the risk of a crisis indicator, the greater the signal error value. As the frequency of risk prediction for a crisis indicator increases, the signal error value decreases. Therefore, the lower the signal error value of the signal indicator, the higher the predictive power of the signal indicator.

Therefore, the signal error value management unit 130 according to the embodiment can select the independent variables with good crisis prediction power through the independent variable NTS step of step 35. [

A description of the signal error value will be further described later with reference to Figs. 6 to 7.

The industrial crisis prediction apparatus 10 according to an exemplary embodiment can determine one integrated crisis index using the independent variables selected in step 35. [ In step 36, the signal error value management unit 130 according to an exemplary embodiment may determine a weight between independent variables to be applied when determining one integrated crisis index through a PCA (Principal Component Analysis) step.

For example, as indicators for forecasting the crisis in the construction industry, four indices can be used, namely, 'market conditions', 'forecasts for funding conditions', 'input forecasts', and 'exchange rates'. All four signal indicators were not missing from the data because the signal error value NTS was lower than the reference value. However, 'market conditions', 'forecasts for funding' and 'input forecasts' are all related to 'construction industry BSI forecast'. If the four signal indicators are given equal weight, the proportion of the 'BSI forecast' in the construction industry is 3 and the 'exchange rate' index is 1, so the forecasting ability of the 'exchange rate' index is underestimated .

Accordingly, in step 37, the signal error value management unit 130 according to the embodiment can treat signal indicators of similar characteristics as a group and correspond to a factor through the PCA step. Accordingly, the signal error value management unit 130 can group a plurality of signal indicators into one or more factors, and treat the grouped factors as one independent variable or one signal indicator.

In step 38, the signal error value management unit 130 can again determine the signal error values for the grouped factors, i.e., one signal index, through the factor calculation step. If there are a plurality of grouped factors, the signal error value NTS is calculated for each signal index at step 35 to analyze the crisis prediction power of the factor by determining the signal error value again for each factor as well as analyzing the crisis prediction power of the signal index .

Since the grouped factors can be regarded as one signal index, the signal error value NTS of the following factor is also referred to as a signal error value NTS of the signal index for the unification of the expressions.

When selecting signal indicators, signal indicators of similar characteristics to be grouped into one factor are not selected at the same time, and only signal indicators of characteristics that are not similar to each other are selected. In this case, May omit the PCA step of step 36 and may not consider the grouped factor.

In step 390, the multi-classification integrated crisis index determination unit 140 may calculate the integrated crisis index of the multi-sector industry group using signal error values NTS determined for each independent variable factor corresponding to the signal index.

As the signal error NTS value is low, the crisis prediction power is a good value, meaning that the larger (1-NTS) value is, the better the crisis prediction power is. Also, taking into account whether or not each factor is a crisis, the value obtained by adding the (1-NTS) values of the factors analyzed as the current crisis state may correspond to the integrated crisis index of the middle class industry.

A description of the composite crisis index of the sub-category industry group will be described in detail later with reference to FIG.

In step 391, the multi-classification integrated crisis index determining unit 140 according to an embodiment can predict the crisis state of the industry using the integrated crisis index of the sub-industry industry group. That is, if a threshold value for indicating the crisis state of the integrated crisis index is set in advance and the integrated crisis index is monitored in accordance with a predetermined cycle, if the integrated crisis index exceeds a predetermined threshold value, a crisis can be predicted have.

6 and 7, a description will be given of a process in which the signal error value management unit 130 according to the embodiment determines a signal error value for one signal indicator.

6 illustrates a chaotic matrix representing a matching relationship between a signal indicator and a crisis indicator in accordance with an embodiment of the present invention.

The chaotic matrix shows the number of cases that can occur in the matching relationship between the leading indicator and the crisis indicator. According to the chaotic process, the number A of the first case in which a crisis signal was generated in the leading indicator and the crisis in the actual crisis index, the number of the second case in which the crisis signal occurred in the leading indicator, , There is a number C in the third case in which the crisis signal did not occur in the leading indicator but the crisis occurred in the actual crisis indicator and a number D in the fourth case in which the crisis signal did not occur in the leading indicator and the crisis did not actually occur in the crisis indicator .

In the first case and the fourth case, it can be said that the crisis was correctly predicted according to the leading indicator. However, in the second case, a false alarm is predicted to predict a crisis in which the leading indicator will not actually occur, and in the third case, a missed call occurs because the leading indicator misses the actual crisis . Therefore, it corresponds to a signal error in the second case and the third case.

The signal error value management unit 130 according to an exemplary embodiment can determine a signal error value NTS that can minimize a signal error in the chaotic matrix. (B / (B + D)) in the second case, and (C / (A + C)) the probability of occurrence of the false alarm failure in the third case. According to one embodiment, the signal error value management unit 130 calculates the probability of occurrence of a crisis signal in a signal indicator before "actual crisis occurs " against" probability of occurrence of a crisis signal in the signal indicator when no actual crisis has occurred &Quot; can be determined as the signal error value NTS. That is, the signal error value NTS may be determined as a ratio of "probability of occurrence of false alarm error" (error probability of second case) to "probability of no false alarm occurrence" (1 - error probability of third case) have.

7 shows a relationship between a signal index and a signal error value according to an embodiment.

In FIG. 6, when a crisis signal of the leading indicator occurs, it means that the leading indicator exceeds the threshold of the leading indicator. Therefore, as the threshold of the leading indicator changes, the numbers in the case of the chaotic matrix can be changed.

FIG. 7 shows a case where the threshold of the unemployment rate is divided into 0.047 threshold intervals from 3.023 to 3.397 when the leading indicator is 'unemployment rate'. It is confirmed that the numbers A, B, C, and D in the first, second, third, and fourth cases fluctuate for each threshold value interval.

The signal error value management unit 130 calculates a signal error value for each threshold interval to determine an optimal threshold value for the preceding indicator, It can be determined as an optimal threshold value for the leading indicator. Also, the signal error value of the minimum value can be determined as the signal error value NTS of the current leading indicator.

Referring to FIG. 7, the signal error value management unit 130 may calculate the signal error value NTS based on the confusion matrix using the corresponding unemployment rate threshold for each threshold interval of the leading indicator 'unemployment rate'. In addition, the signal error value management unit 130 can also calculate the predictive power (A / (A + B)) that indicates the rate at which the crisis actually occurred in the crisis index when the crisis signal was generated in the preceding indicator.

The signal error value management unit 130 according to the embodiment derives the signal error value NTS of the minimum value among the plurality of threshold sections and outputs the threshold value at which the prediction power of the maximum value is derived, It can be determined to be a low threshold value. Therefore, the optimal threshold for the leading indicator 'unemployment rate' is determined to be 3.677, and the signal error value NTS for the leading indicator 'unemployment rate' can be determined to be 0.20.

The process of determining the signal error value NTS for one signal index has been described above in the signal error value management unit 130 according to the embodiment. Since the signal index management unit 120 according to an exemplary embodiment selects a plurality of signal indicators preceding the one risk indicator, the signal error value management unit 140 can determine the signal error value NTS for each signal indicator.

The multi-classification integrated crisis index determining unit 140 according to an embodiment can determine the multi-classification integrated crisis index using the signal error values of the signal indicators for the crisis indicators of the current classification industry group.

Where i represents the identification number of a plurality of signal indicators, NTS _i represents the error signal value of the i-th signal NTS surface. F _i represents a value of 1 if the i th signal indicator is a crisis signal, or 0 otherwise. As the signal error value NTS itself is smaller, the actual crisis prediction power of the corresponding signal indicator is good, and (1-NTS _i ) means that the larger the value of the signal error value NTS itself, the better the actual crisis prediction power of the signal indicator. Therefore, the integrated classification crisis index is a value calculated by reflecting the ratio of actual crisis prediction power to whether the signal indicator is a crisis signal at the time of judgment.

FIG. 8 illustrates the leading and window periods of a signal indicator and a crisis indicator according to an embodiment.

Window period means a specific sample time for determining a threshold value of a signal index. For example, if the window period 95 of the signal indicator begins to be collected from May 2006, the window period 90 of the crisis indicator that tests whether an actual crisis occurs is equal to or greater than the window period 95 of the signal indicator Month, starting in November 2006. Also, if the signal index management unit 120 examines whether a real crisis occurs for a predetermined period of two years, three years, or five years in the window period 90 of the crisis index, the window period 95 of the signal index Signal indicators can be collected for the same period from the start time.

The signal error value management unit 130 according to an embodiment may collect the result of the crisis indicators for the signal indicators preceding 6 months during the window period to calculate a threshold value of the signal indicator that becomes the minimum signal error value NTS .

The industrial crisis prediction apparatus 100 according to the embodiment can predict the actual crisis indicator after the window period 90 of the crisis indicator completes. If a signal indicator occurs at a threshold value or more of the signal indicator, Can be interpreted as having occurred.

FIG. 9 shows a graph 60 of the integrated crisis index of the multiple classification industry group according to one embodiment.

Figure 9 shows a graph of the composite crisis index (60) of the signal indicators for the trend of non-performing loans in the classification category. The multi-class aggregate crisis index determining unit 140 according to the embodiment determines the sub-aggregate crisis index using the signal error values of the signal indicators for the trend of the non-current loan credit per month, 60 can be derived.

The signal indicator threshold 65 corresponds to the minimum signal error value as described above with reference to FIG. The industrial crisis prediction apparatus 100 according to an embodiment determines the integrated crisis index of the middle class industry group and then calculates the integrated crisis index based on the signal index threshold value 65 based on the stability index, the surrounding interval, the crisis interval, and the danger interval Can be analyzed. The current crisis situation can be expected depending on which segment the sub-category comprehensive crisis index belongs to.

Accordingly, the industrial group crisis prediction apparatus 100 can analyze an interval exceeding the signal index threshold 65 in the integrated crisis index graph 60 of the signal indicators into the interval 63 in which the crisis signal according to the signal indicator occurs have. In addition, the interval 63 in which a crisis signal according to the signal index occurs can be interpreted as a crisis period or a dangerous period.

FIG. 10 shows a graph of a total crisis index 60 and a trend graph 80 of the proportion of non-interest loans in the multiple classification industry group according to an embodiment.

According to the industrial military crisis prediction apparatus 100 according to an embodiment, when a period 63 in which a crisis signal according to a signal indicator is generated is monitored, a crisis may also be expected to occur in the actual crisis indicator of the secondary classification industry.

For example, the dependent variable may be the ratio of non-performing loans to the rate of delinquency. The multi-classification comprehensive crisis index determining unit 140 according to an embodiment can determine the threshold value (85) of the non-interest loaned portion as a dependent variable according to the following relational expression.

Here, it can be seen that a real crisis has occurred in the section 83 where the trend graph 80 of non-interest loans has exceeded the threshold value 85 of the non-interest loan portion, and the actual crisis occurs.

A period 63 in which a crisis signal according to signal indicators is generated in the middle classification total crisis index graph 60 determined by the middle class comprehensive crisis index determiner 140 according to an embodiment and a transition graph of non- ), A certain time lag occurs between the intervals 83 in which actual crisis occurs. That is, when the interval 63 in which the crisis signal is generated according to the signal indicators is observed, the middle class comprehensive crisis index determination unit 140 may be expected to have a period 83 in which a real crisis occurs after a certain time lag .

The process of predicting the crisis situation of each sub-industry group using the sub-classification comprehensive crisis index has been described above with reference to FIGS. 9 and 10.

11 illustrates a candidate set of signal indicators according to one embodiment.

11 shows a candidate group of a leading indicator and signal indicators that can be used as an independent variable in the industrial military crisis prediction apparatus 100 according to an exemplary embodiment. The signal indicators are divided into macroeconomic indicators, industry trends, and financial indicators. The subdivided signal indicators can be collected according to their characteristics, such as quarter, month, and the like.

In the industrial crisis prediction apparatus 100 according to an exemplary embodiment, signal indicators having a low signal error value and a high crisis prediction ability for the current industry group among the candidate signals of the signal indicators can be selected according to the characteristics of the industry group. In addition, signal indicators with low similarity can be finally selected.

It should be noted that the indicators shown in FIG. 11 are merely examples, but the economic indicators applicable to the industrial crisis prediction apparatus 100 according to the present invention are not limited to those shown in FIG.

In one embodiment, the index database 110, the signal index management unit 120, the signal error value management unit 130, and the middle classification crisis index determining unit 140 of the industrial military crisis prediction apparatus 100 operate in accordance with individual cycles .

For example, the signal index management unit 120 can update the index database every six months by selecting signal indicators for the crisis index and storing the index database 100. The signal threshold value management unit 130 according to the embodiment can update the signal error value of the independent variable every 6 months and the threshold value of the signal error value at which the signal error value becomes the lowest can be updated every 6 months. The information on the directionality of the independent variable indicating whether the signal index at the current time point is in the state of rising or falling compared to the threshold value can also be updated every six months.

For example, the multi-segment comprehensive crisis index determination unit 140 may analyze the current signal error value of the signal indicator every month to monitor whether each independent variable generates a crisis signal. Also, using the signal error value of the signal indicators analyzed every month and the threshold value, the integrated classification crisis index can be updated every month.

The block diagrams disclosed herein may be construed to those skilled in the art to conceptually represent circuitry for implementing the principles of the present invention. Likewise, any flow chart, flow diagram, state transitions, pseudo code, etc., may be substantially represented in a computer-readable medium to provide a variety of different ways in which a computer or processor, whether explicitly shown or not, It will be appreciated by those skilled in the art. Therefore, the above-described embodiments of the present invention can be realized in a general-purpose digital computer that can be created as a program that can be executed by a computer and operates the program using a computer-readable recording medium. The computer-readable recording medium includes a storage medium such as a magnetic storage medium (e.g., ROM, floppy disk, hard disk, etc.), optical reading medium (e.g., CD ROM,

The functions of the various elements shown in the figures may be provided through use of dedicated hardware as well as hardware capable of executing the software in association with the appropriate software. When provided by a processor, such functionality may be provided by a single dedicated processor, a single shared processor, or a plurality of individual processors, some of which may be shared. Also, the explicit use of the term " control portion "or" portion "should not be construed to refer exclusively to hardware capable of executing software and includes, without limitation, digital signal processor (DSP) hardware, (ROM), a random access memory (RAM), and a non-volatile storage device.

In the claims hereof, the elements depicted as means for performing a particular function encompass any way of performing a particular function, such elements being intended to encompass a combination of circuit elements that perform a particular function, Or any form of software, including firmware, microcode, etc., in combination with circuitry suitable for carrying out the software for the processor.

Reference throughout this specification to " one embodiment " of the principles of the invention and various modifications of such expression in connection with this embodiment means that a particular feature, structure, characteristic or the like is included in at least one embodiment of the principles of the invention it means. Thus, the appearances of the phrase " in one embodiment " and any other variation disclosed throughout this specification are not necessarily all referring to the same embodiment.

In this specification, the expression 'at least one of' in the case of 'at least one of A and B' means that only the selection of the first option (A) or only the selection of the second listed option (B) It is used to encompass the selection of options (A and B). As an additional example, in the case of 'at least one of A, B and C', only the selection of the first enumerated option (A) or only the selection of the second enumerated option (B) Only the selection of the first and second listed options A and B or only the selection of the second and third listed options B and C or the selection of all three options A, B, and C). Even if more items are listed, they can be clearly extended to those skilled in the art.

The present invention has been described with reference to the preferred embodiments.

It is to be understood that all embodiments and conditional statements disclosed herein are intended to assist the reader in understanding the principles and concepts of the present invention to those skilled in the art, It will be understood that the invention may be embodied in various other forms without departing from the spirit or essential characteristics thereof. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

Claims (10)

In a crisis forecasting method for preemptive risk management,
Storing, in a database, a plurality of signal indicators selected to explain a crisis state for each of the subdivided industry groups for each subdivided plurality of subdivided industry groups;
Calculating a signal error value indicating a degree of each signal indicator correctly indicating an actual crisis state of a corresponding industry group and a threshold value of the signal error value for each of the signal indicators; And
Determining an integrated crisis index of each of the multiple classification industry groups on the basis of signal error values of the signal indicators and whether the signal error values exceed a corresponding threshold value for each of the secondary classification industry groups. Prediction method. 2. The method of claim 1, wherein the step of calculating a threshold value of the signal error value and the signal error value comprises:
Determining a first signal indicator among the signal indicators and a first crisis indicator caused by the first signal indicator;
Determining the signal error value indicating a probability that the first crisis indicator corresponding to the first signal indicator incorrectly represents the actual crisis state of the industry during the window period from the time of occurrence of the first crisis indicator; And
And determining a threshold of the first signal indicator that minimizes the signal error value. 3. The method of claim 2, wherein determining the signal error value comprises:
From the first signal indicator at a time point preceding the first crisis indicator to the first signal indicator during the window period from the generation time point of the first crisis indicator, Analyzing whether the crisis indicator represents an actual crisis state;
During the window period, a ratio of a first error in which the actual crisis state did not occur but a crisis occurred in the first crisis indicator and a second error in which the actual crisis state occurred but the crisis did not occur in the first crisis indicator, Determining the signal error value using the signal error value; And
And determining a threshold of the first signal indicator that minimizes the signal error value. The method according to claim 1, wherein the step of determining the total crisis index of each of the sub-
Comparing the signal error value with a corresponding threshold value individually determined for each signal error value and summing at least one signal error value exceeding the corresponding threshold value; And
And determining a negative value of the summed signal error value as a value corresponding to the integrated crisis index of the sub-category industry group. 4. The method according to claim 3,
Further comprising the step of monitoring a crisis state for said middle classification industry group when the total crisis index of said middle classification industry group exceeds a threshold value. In an industry crisis prediction apparatus for preemptive risk management,
A database for storing various indicator data for predicting industrial crisis for preemptive risk management;
A signal indicator managing unit for periodically selecting a plurality of signal indicators for describing a crisis state for each of the plurality of classified industry groups classified by the major classification industry group and storing the signal indicators in the database;
A signal error value indicating a degree of correctness of each signal indicator of an actual industry in each of the signal indicators; a signal error value management unit for calculating a threshold value of the signal error value and storing the signal in the database; And
And a middle classification total crisis index determining unit for determining a total crisis index of each of the multiple classification industry groups based on signal error values of the signal indicators and whether the signal error values exceed a corresponding threshold value The industrial crisis prediction device. 7. The apparatus of claim 6, wherein the signal error value management unit comprises:
Determining a first signal indicator among the signal indicators and a first crisis indicator generated by the first signal indicator,
Determining the signal error value indicating a probability that the first crisis indicator corresponding to the first signal indicator incorrectly indicates an actual crisis state of the industry during the window period from the time of occurrence of the first crisis indicator,
And determines a threshold value of the first signal indicator that minimizes the signal error value. The apparatus of claim 7, wherein the signal error value management unit comprises:
A signal error value which is individually determined for each of the signal error values is compared with the signal error value to sum up at least one signal error value exceeding the corresponding threshold value and a negative value of the summed signal error value And determining a value corresponding to the total crisis index as a value corresponding to the total crisis index. The system according to claim 6, wherein the multi-
And compares the signal error value with a corresponding threshold value individually determined for each of the signal error values to add at least one signal error value exceeding the corresponding threshold value to a negative value of the summed signal error value, And determining a value corresponding to the total crisis index as a value corresponding to the total crisis index. The industrial crisis prediction apparatus according to claim 9,
Further comprising a real-time monitoring unit for monitoring a crisis state of the classified industry group when the integrated crisis index of the classified industry group exceeds a threshold value.

Images