Green bond index prediction method and device

Abstract

The invention belongs to the technical field of green bond prediction, and provides a green bond index logarithmic revenue prediction method and device, and the method comprises the steps: constructing influence factor variables dynamically related to a green bond market, and the influence factor variables comprise at least one group of influence factor variables; influence factor variables and green bond index logarithmic revenue in a specified historical time period are collected, a training data set is established, and the green bond index logarithmic revenue is a time sequence value obtained after logarithmic transformation differential processing is carried out on the time dimension; based on a quantile adaptive group Lasso estimator, performing parameter estimation by adopting the established training data set so as to construct a green bond index estimator; and predicting the green bond index logarithmic revenue by adopting the constructed green bond index estimator according to a parameter estimation value obtained by parameter estimation. According to the method, the dynamic variation of the green bond index in the specified time period can be accurately predicted.

Description

Technical Field

[0001] The present invention relates to the technical field of green bond prediction, and in particular to a green bond index prediction method and device. Background Art

[0002] In recent years, environmental issues have attracted widespread attention around the world. As an important measure to address energy challenges, climate change and sustainable development, green bonds have quickly become a relatively new financial instrument and have played an important role. Green bonds significantly improve resource efficiency by directing funds to environmental protection fields such as renewable energy, energy efficiency improvement, clean transportation and sustainable buildings (Gupta, 2017; Ng and Tao, 2016). Therefore, they play a vital role in reducing carbon emissions, protecting ecosystems and promoting low-carbon economic development (Flaherty et al., 2017; Reboredo and Ugolini, 2020). The first green bond was issued by the European Investment Bank in 2007 for environmental protection projects (Gianfrate and Peri, 2019). In the decade since its emergence, the international green bond market has grown rapidly. According to the Climate Bonds Initiative (CBI), green bond issuance has increased from US$36.6 billion in 2014 to US$517.4 billion in 2021. The significant growth of the green bond market has been described as the "green bond boom" (Jin et al., 2020). As the number of green bond issuances and investment options increase, a tool is needed to measure and compare different bonds. In response to this demand, green bond indices have emerged as an important tool for evaluating the performance of green bonds. They provide greater transparency, credibility and simplified risk assessment (Baltas and Mann, 2023). The S&P Green Bond Index, launched by Standard & Poor's Global Ratings on July 31, 2014, is widely recognized as the most commonly used index (Yaya et al., 2022; Abakah et al., 2022). It is the first index to track the global green bond market (Kochetygova and Jauhari, 2014). By monitoring the S&P Green Bond Index, future trends in the green bond market can be predicted and appropriate adjustments can be made to its investment portfolio.

[0003] Predicting green bond indices is a challenging task. First, research on the key factors for prediction is scarce, resulting in limited information. Second, the model suitable for this task has not yet been determined. In fact, the return data of green bond indices do not follow a normal distribution, which makes many forecasting models that rely on the normality assumption inapplicable.

[0004] However, due to the particularity of the green bond market and data characteristics, existing methods still face many challenges in their application: (1) they mainly focus on mean prediction and fail to capture the tail behavior and asymmetry of the green bond index; (2) they only focus on a small number of variables and fail to fully utilize a large number of potential predictive factors; (3) even if multiple predictive factors are considered, the rich information in high-dimensional data cannot be fully utilized due to the limited application of variable selection methods.

[0005] Therefore, it is necessary to provide a new green bond index prediction method and device to solve the above problems. Summary of the invention

[0006] The present invention aims to provide a green bond index prediction method and device to solve the technical problems in the prior art that the existing methods fail to make full use of a large number of potential prediction factors, resulting in low prediction accuracy, fail to make full use of rich information data in high-dimensional data, resulting in low prediction accuracy, and the prediction model that relies on the normality assumption is not applicable because the return data of the green bond index does not follow the normal distribution. The technical problem to be solved by the present invention is achieved by the following technical solution.

[0007] The first aspect of the present invention provides a green bond index prediction method, comprising: constructing influencing factor variables related to the green bond market dynamics, wherein the influencing factor variables include at least one group of influencing factor variables; collecting the influencing factor variables and the logarithmic returns of the green bond index within a specified historical time period to establish a training data set, wherein the logarithmic returns of the green bond index are time series values ​​obtained after logarithmic transformation and difference processing in the time dimension; based on the quantile adaptive group Lasso estimator, using the established training data set to perform parameter estimation to construct a green bond index estimator; according to the parameter estimation values ​​obtained by the parameter estimation, using the constructed green bond index estimator to predict the dynamic change of the green bond index.

[0008] The second aspect of the present invention provides a green bond index prediction device, which executes the green bond index prediction method described in the first aspect of the present invention, and the green bond index prediction device includes: a construction module, which is used to construct influencing factor variables related to the dynamics of the green bond market, and the influencing factor variables include at least one group of influencing factor variables; a collection and processing module, which is used to collect influencing factor variables and green bond index logarithmic returns within a specified historical time period, and establish a training data set, and the green bond index logarithmic returns are time series values ​​obtained after logarithmic transformation and difference processing in the time dimension; a construction module, based on the quantile adaptive group Lasso estimator, using the established training data set to perform parameter estimation to construct a green bond index estimator; a prediction module, according to the parameter estimation value obtained by the parameter estimation, using the constructed green bond index estimator, predicting the dynamic change of the green bond index. The third aspect of the present invention provides an electronic device, including: one or more processors; a storage device, which is used to store one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the green bond index prediction method described in the first aspect of the present invention.

[0009] The fourth aspect of the present invention provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the green bond index prediction method described in the first aspect of the present invention.

[0010] Beneficial effects of the present invention:

[0011] Compared with the prior art, the present invention is based on constructing influencing factor variables related to the dynamics of the green bond market; collecting influencing factor variables and green bond index logarithmic returns within a specified historical time period to establish a training data set, wherein the green bond index logarithmic return is a time series value obtained after logarithmic transformation and difference processing on the time dimension; based on the quantile adaptive group Lasso estimator, the established training data set is used for parameter estimation to construct a green bond index estimator; according to the parameter estimation values ​​obtained by the parameter estimation, the constructed green bond index estimator is used to accurately predict the dynamic changes of the green bond index; using multiple groups of influencing factor variables in high-dimensional data, and performing data analysis from the time dimension, the parameter estimation values ​​are adjusted, that is, the self-constructed green bond index estimator is used to accurately predict the value of the green bond index and its dynamic changes within the specified time period. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a flowchart of an example of the green bond index prediction method of the present invention;

[0013] Figure 2It is a partial flow chart of a specific implementation method of the green bond index prediction method of the present invention;

[0014] Figure 3 is a schematic diagram of an example of a green bond index prediction device of the present invention;

[0015] Figure 4 is a schematic structural diagram of an electronic device embodiment according to the present invention;

[0016] Figure 5 It is a schematic diagram of the structure of a computer-readable medium embodiment according to the present invention. DETAILED DESCRIPTION

[0017] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0018] In view of the above problems, the present invention provides a green bond index prediction method, which constructs a green bond index estimator; based on the parameter estimation values ​​obtained by parameter estimation, the constructed green bond index estimator is used to predict the dynamic changes of the green bond index; multiple groups of influencing factor variables in high-dimensional data are used, and data analysis is performed from the time dimension, and the parameter estimation values ​​are adjusted, that is, a self-constructed green bond index estimator is used to predict the value of the green bond index and its dynamic changes within a specified time period.

[0019] Specifically, we use the quantile-based data dimensionality reduction method to solve the problem of predicting green bond market returns. We construct a green bond index estimator through a quantile adaptive dimensionality reduction model. Based on the two significant properties of the proposed method, namely "interpretability of the final estimator" and "fast calculation", we select multiple groups of influencing factor variables from a comprehensive pool related to the dynamics of the green bond market and identify the most powerful predictors to predict the dynamic changes in future green bond returns, while allowing for potential differences in significant predictors at different green bond return quantiles.

[0020] Example 1

[0021] Refer to the following Figure 1 , Figure 2 , the method of the present invention will be described in detail.

[0022] Figure 1 The figure is a flowchart of an example of the green bond index prediction method of the present invention.

[0023] First, in step S101, influencing factor variables related to the green bond market dynamics are constructed, and the influencing factor variables include at least one group of influencing factor variables.

[0024] Specifically, the following groups of influencing factor variables are used to construct influencing factor variables related to the dynamics of the green bond market: energy source group, energy price group, stock market index group, monetary policy group and economic information group.

[0025] In a specific implementation, the Standard & Poor's (S&P) Green Bond Index is used as an example. The green bond index is designed to track the global green bond market. Since the green bond index fluctuates continuously during trading hours, in this implementation, the prediction factor (i.e., the dependent variable) is the logarithmic return of the green bond index, that is, the green bond index return rate is defined as the logarithmic monthly change rate of the Standard & Poor's Green Bond Index data. Specifically, the green bond index data within a specific time period is logarithmically transformed and the logarithm is taken. For example, the green bond index at time t0 and time t is respectively denoted as p t-1 and p t , the logarithmic return of the green bond index at time t is defined as log(p t )-log(p t-1 ).

[0026] At least one group of influencing factor variables is used. In this embodiment, the following five groups of influencing factor variables are used to predict the return rate of the green bond index: energy source group, energy price group, stock market index group, monetary policy group and economic information group. In order to ensure stability, these influencing factor variables need to be preprocessed by logarithmic transformation and difference transformation, that is, logarithmic return rate.

[0027] Optionally, the influencing factor variables also include at least one of the following economic variables: crude oil related parameters, natural gas related parameters, natural gas prices and crude oil prices in a designated country, and relevant indexes in a designated country.

[0028] Specifically, the influencing variables of the energy source group include crude oil production, crude oil imports, crude oil inventories, natural gas imports and natural gas production. The influencing variables of the energy price group include the spot price of crude oil in the specified country, the spot price of crude oil (WTI) in the specified country, the natural gas futures price of the exchange (ICE), the natural gas futures price of the commodity exchange (NYMEX) and the natural gas futures price of the energy exchange (European Energy Exchange).

[0029] The influencing variables of the stock market index group include the industrial average index of the specified country, CAC 40 index, DAX index, TSX index, MIB index, BOVESPA index, MOEX index, etc. The influencing variables of the monetary policy group include G7 and the money supply M2 of the specified country. The influencing variables of the economic information group include the consumer price index, unemployment rate, and long-term bond yields.

[0030] Comprehensively considering the variables that affect the green bond index, specifically the influencing factor variable package of energy source group, energy price group, stock market index group, monetary policy group and economic information group, can significantly improve the prediction accuracy.

[0031] It should be noted that in other examples, two, three, four, six, or more groups of influencing factor variables may also be used. The above is only described as an optional example and cannot be understood as a limitation of the present invention.

[0032] Next, in step S102, the influencing factor variables and the logarithmic returns of the green bond index within a specified historical time period are collected to establish a training data set. The logarithmic returns of the green bond index are time series values ​​obtained after logarithmic transformation and difference processing in the time dimension.

[0033] The influencing factor variables and green bond index collected within the specified historical time period are used to establish a training data set and a test data set, and the training data set is used for parameter estimation. The test data set is used for testing to obtain a prediction result, and the prediction result is compared with the true value to further adjust and optimize the parameter estimation value. The specified historical time period (specifically corresponding to the historical time end corresponding to the initial time t0 to the time t below) includes six months, ten months, twelve months, twenty-four months, etc.

[0034] Specifically, for example, using the observations of the monthly S&P Green Bond Index from January 2012 to December 2022, a predictor, namely the S&P Green Bond Index Log Return, is constructed, which is defined as the logarithmic monthly change of the S&P Green Bond Index.

[0035] For example, the sample data from January 2012 to December 2021 is used as the samples in the training data set to form a training data set, and the sample data from January 2022 to December 2022 is used as the samples in the test data set to form a test data set. Through this time dimension segmentation method, the prediction ability of the model (i.e., the constructed green bond index estimator) is evaluated from the time dimension, so as to better optimize the relevant parameters of the green bond index estimator and achieve higher prediction accuracy.

[0036] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0037] Next, in step S103, based on the quantile adaptive group Lasso estimator, the established training data set is used to perform parameter estimation to construct a green bond index estimator.

[0038] Specifically, based on the quantile adaptive group Lasso estimator, the established training data set is used for parameter estimation.

[0039] It should be noted that the quantile adaptive dimensionality reduction model is built on the quantile regression model. Quantile regression focuses on obtaining the conditional median or conditional quantile information of the response variable. Compared with the standard linear regression that can only capture the relationship between the predictor variable and the mean response, quantile regression can provide more information about the different conditional quantiles of the response variable. Therefore, the influence of outliers in the analysis is smaller.

[0040] Adaptive group LASSO penalty function is used. Assume that the regression coefficient vector β has a group structure, {g1,g2,...,g q} is a combination of index numbers {1,2...,n}, where n is the number of predictor variables. and (s represents the group index, ranging from 1 to q) then the regression coefficient vector in, m and n are both positive integers. In the quantile regression with adaptive group LASSO penalty, parameter estimation is defined.

[0041] Specifically, the following expressions are used to estimate the parameters and obtain the updated parameter estimates:

[0042]

[0043] in, are the parameter estimates, which is a vector where represents the intercept parameter estimate of the quantile adaptive group Lasso model obtained using historical data from time t0 to time t. It represents the estimated value of the regression vector of the quantile adaptive group Lasso model obtained using the historical data from time t0 to time t. Specifically, it is a multidimensional vector, including multiple parameters; α is the intercept parameter, β represents the regression coefficient vector, and λ is the penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, β i is the i-th parameter in the regression coefficient vector β, and the i-th parameter is related to the predicted variable p i,j Correspondingly, the value range of i is 1 to n, where n is the number of predictor variables, and the regression coefficient vector β has a group structure, that is, q is the number of groups, k is the group length index, is the regression coefficient vector β in g k The parameters in the group, and Among them, T k is the regression coefficient vector βg k The number of parameters in the group, is the gth in the regression coefficient vector β k The lth value in the group, Indicates g k The adaptive weight of the group.

[0044] Optionally, the following multiple groups of influencing factor variables are used to construct influencing factor variables related to the dynamics of the green bond market: energy source group, energy price group, stock market index group, monetary policy group and economic information group.

[0045] The penalty item parameter is selected based on the cross-validation method or the BIC value, and the value range of the penalty item parameter λ is greater than 0.

[0046] like Figure 2 As shown, g k Adaptive weights for parameters in a group The steps to determine the initial value of are:

[0047] Step S210: Using the quantile group Lasso model, obtain The parameter estimate vector for .

[0048] The specific calculation is done using the following expression:

[0049]

[0050] in, are the initial parameter estimates, where It represents the initial intercept parameter estimate obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. It represents the estimated value of the initial regression vector obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. a is the initial intercept parameter, b represents the initial regression coefficient vector, and γ is the initial penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, b i is the i-th parameter in the initial regression coefficient vector b. The i-th parameter is related to the predicted variable p i,jCorrespondingly, the value range of i is 1 to n, n is the number of predictor variables, q is the number of groups, and k is the group length indicator. After grouping the initial regression coefficient vector b, g k The parameters in the group, and g represents the initial regression coefficient vector b k The hth parameter in the group, h=1,…,Z k , where Z k is the initial regression coefficient vector bg k The number of parameters in the group, u k Yes k The penalty weight of the group is set

[0051] Step S220: Further based on the obtained parameter estimation value, the adaptive weight is calculated using the following expression:

[0052]

[0053] in

[0054]

[0055] is the gth estimated value of the regression vector obtained in step S210 k The value of the group, is the gth regression vector estimate k The hth value in the group.

[0056] Compared with traditional dimensionality reduction methods (such as Lasso), which often focus on mean prediction and are therefore unable to adapt to the different tail behaviors and asymmetry of green bond index data, the present invention can combine the quantile regression model with the adaptive group Lasso penalty term to construct a quantile adaptive group Lasso prediction model, so that it can better process green bond index data. Moreover, compared to the classic Lasso model, the adaptive group Lasso penalty term takes into account the common situation of group structure in the features. In this case, the grouping structure contains additional information, so ignoring it and using standard methods will result in inaccurate estimators. At this point, the adaptive group Lasso can collect more complex structural information between variables, so it can make the prediction results more accurate.

[0057] Using the quantile group lasso model to obtain the parameter g k Initial values ​​of adaptive weights in the group Based on the initial values ​​obtained in step S210, using the quantile group adaptive lasso estimator, variables can be effectively selected and estimated in a high-dimensional data environment.

[0058] Specifically, quantile regression analysis is performed on the influencing factor variables to obtain the conditional median or conditional quantile information of the response variable. By introducing an adaptive group Lasso penalty term in the quantile regression analysis, variables can be effectively selected and estimated in a high-dimensional data environment.

[0059] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0060] Next, in step S104, the constructed green bond index estimator is used to predict the dynamic change of the green bond index based on the parameter estimation values ​​obtained by parameter estimation.

[0061] Specifically, the following expression is used to predict the value of the green bond index:

[0062]

[0063] in, is the predicted value of the logarithmic return of the green bond index at time t+1, represents the intercept parameter estimate based on the historical data from time t0 to time t, represents the estimated value of the i-th regression parameter based on the historical data from time t0 to time t, p i,t is the i-th predictor variable at time t, the value of i ranges from 1 to n, and n is the number of predictor variables.

[0064] For example, we use the monthly observations of the S&P Green Bond Index from January 2012 to December 2022 to construct a predictor, the S&P Green Bond Index Log Return, which is defined as the log monthly change in the S&P Green Bond Index.

[0065] Further calculate the dynamic changes of the green bond index from the current moment to the future six months, ten months, twelve months, and twenty-four months.

[0066] Specifically, the dynamic variables for predicting the future green bond index include predicting changes in the logarithmic returns of the green bond index.

[0067] In an optional implementation, the dynamic variables of the future green bond index include predicting the change in the median of the logarithmic return of the green bond index. The median is selected as a value including the median, and the value is in the range of 0.1 to 1.0. Specifically, the value of the median is 0.1, 0.3, 0.5 (median), 0.7, 0.9, etc.

[0068] It should be noted that the above is only described as an optional example and should not be understood as a limitation to the present invention.

[0069] Compared with the prior art, the present invention is based on constructing influencing factor variables related to the dynamics of the green bond market; collecting influencing factor variables and green bond index logarithmic returns within a specified historical time period to establish a training data set, wherein the green bond index logarithmic return is a time series value obtained after logarithmic transformation and difference processing on the time dimension; based on the quantile adaptive group Lasso estimator, the established training data set is used for parameter estimation to construct a green bond index estimator; according to the parameter estimation values ​​obtained by the parameter estimation, the constructed green bond index estimator is used to predict the dynamic changes of the green bond index; using multiple groups of influencing factor variables in high-dimensional data, and performing data analysis from the time dimension, the parameter estimation values ​​are adjusted, that is, the self-constructed green bond index estimator is used to predict the value of the green bond index and its dynamic changes within a specified time period.

[0070] Example 2

[0071] The following are embodiments of the device of the present invention, which can be used to implement the method embodiments of the present invention. For details not disclosed in the device embodiments of the present invention, please refer to the method embodiments of the present invention.

[0072] Figure 3 4 is a schematic structural diagram of an example of a green bond index prediction device according to the present invention.

[0073] The following will refer to Figure 3 , a green bond index prediction device 300 is described, which executes the green bond index prediction method described in the present invention.

[0074] The green bond index prediction device 300 includes a construction module 310 , a collection and processing module 320 , a building module 330 , and a prediction module 340 .

[0075] In a specific embodiment, the construction module 310 is used to construct influencing factor variables related to the dynamics of the green bond market, and the influencing factor variables include at least one group of influencing factor variables. The collection and processing module 320 is used to collect the influencing factor variables and the logarithmic returns of the green bond index within a specified historical time period to establish a training data set, and the logarithmic returns of the green bond index are time series values ​​obtained after logarithmic transformation and difference processing in the time dimension. The construction module 330 uses the established training data set to perform parameter estimation based on the quantile adaptive group Lasso estimator to construct a green bond index estimator. The prediction module 340 uses the constructed green bond index estimator to predict the dynamic changes of the green bond index based on the parameter estimation values ​​obtained by the parameter estimation.

[0076] According to an optional embodiment, the quantile-based adaptive group Lasso estimator uses the established training data set to perform parameter estimation to construct a green bond index estimator.

[0077] Use the following expression to estimate the parameters and obtain the updated parameter estimates:

[0078]

[0079] in, are the parameter estimates, which is a vector where represents the intercept parameter estimate of the quantile adaptive group Lasso model obtained using historical data from time t0 to time t. It represents the estimated value of the regression vector of the quantile adaptive group Lasso model obtained using the historical data from time t0 to time t. Specifically, it is a multidimensional vector, including multiple parameters; α is the intercept parameter, β represents the regression coefficient vector, and λ is the penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, βi is the i-th parameter in the regression coefficient vector β, and the i-th parameter is related to the predictor variable p i,j Correspondingly, the value range of i is 1 to n, where n is the number of predictor variables, and the regression coefficient vector β has a group structure, that is, q is the number of groups, k is the group length index, is the regression coefficient vector β in g k The parameters in the group, and Among them, T k is the regression coefficient vector βg k The number of parameters in the group, is the gth in the regression coefficient vector β k The lth value in the group, Indicates g k The adaptive weight of the group.

[0080] The following groups of influencing factor variables are used to construct influencing factor variables related to the dynamics of the green bond market: energy source group, energy price group, stock market index group, monetary policy group and economic information group.

[0081] According to an optional embodiment, the quantile-based adaptive group Lasso estimator uses the established training data set to perform parameter estimation to construct a green bond index estimator.

[0082] The penalty item parameter is selected based on the cross-validation method or the BIC value, and the value range of the penalty item parameter λ is greater than 0.

[0083] g k Adaptive weights for parameters in a group The steps to determine the initial value of are:

[0084] Step S210: using the quantile group Lasso model to obtain a parameter estimation vector;

[0085]

[0086] in, are the initial parameter estimates, where It represents the initial intercept parameter estimate obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. It represents the estimated value of the initial regression vector obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. a is the initial intercept parameter, b represents the initial regression coefficient vector, and γ is the initial penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, b i is the i-th parameter in the initial regression coefficient vector b. The i-th parameter is related to the predicted variable p i,j Correspondingly, the value range of i is 1 to n, n is the number of predictor variables, q is the number of groups, and k is the group length indicator. After grouping the initial regression coefficient vector b, g k The parameters in the group, and g represents the initial regression coefficient vector b k The hth parameter in the group, h=1,…,Z k , where Z k is the initial regression coefficient vector bg k The number of parameters in the group, u k Yes k The penalty weight of the group is set

[0087] Step S220: Further based on the obtained parameter estimation value, the adaptive weight is calculated using the following expression:

[0088]

[0089] in

[0090]

[0091] is the gth estimated value of the regression vector obtained in step S210 k The value of the group, is the gth regression vector estimate k The hth value in the group.

[0092] According to an optional implementation manner, the parameter estimation values ​​obtained according to the parameter estimation are used to predict the dynamic changes of the green bond index using the constructed green bond index estimator.

[0093] Specifically, the following expression is used to predict the value of the logarithmic return of the green bond index:

[0094]

[0095] in, is the predicted value of the logarithmic return of the green bond index at time t+1, represents the intercept parameter estimate based on the historical data from time t0 to time t, represents the estimated value of the i-th regression parameter based on the historical data from time t0 to time t. The estimated value of the i-th regression parameter is related to the predicted variable p i,j Correspondingly, p i,t is the i-th predictor variable at time t, the value of i ranges from 1 to n, and n is the number of predictor variables.

[0096] According to an optional implementation manner, the influencing factor variables also include at least one of the following economic variables: crude oil related parameters, natural gas related parameters, natural gas prices and crude oil prices in a designated country, and relevant indexes of a designated country.

[0097] According to an optional implementation, the dynamic variables for predicting the future green bond index include predicting changes in the logarithmic returns of the green bond index, wherein the median is selected as a value including the median, and the value is in the range of 0.1 to 1.0.

[0098] According to an optional implementation manner, the influencing factor variables and green bond index collected within a specified historical time period are used to establish a training data set and a test data set, and the training data set is used to perform parameter estimation to obtain parameter estimation values; the obtained parameter estimation values ​​are used to perform testing using the test data set to obtain prediction results, and the prediction results are compared with the true values ​​to further adjust and optimize the parameter estimation values.

[0099] It should be noted that due to Figure 3Green bond index prediction method and green bond index prediction device Figure 1 The green bond index prediction methods in the examples are roughly the same, so the description of the same parts is omitted.

[0100] Compared with the prior art, the present invention is based on constructing influencing factor variables related to the dynamics of the green bond market; collecting influencing factor variables and green bond index logarithmic returns within a specified historical time period to establish a training data set, wherein the green bond index is a time series value obtained after logarithmic transformation and difference processing on the time dimension; based on the quantile adaptive group Lasso estimator, the established training data set is used for parameter estimation to construct a green bond index estimator; according to the parameter estimation values ​​obtained by the parameter estimation, the constructed green bond index estimator is used to predict the dynamic changes of the green bond index; using multiple groups of influencing factor variables in high-dimensional data, and performing data analysis from the time dimension, the parameter estimation values ​​are adjusted, that is, the self-constructed green bond index estimator is used to predict the value of the green bond index and its dynamic changes within a specified time period.

[0101] Example 3

[0102] Figure 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

[0103] like Figure 4 As shown, the electronic device is presented in the form of a general computing device. The processor may be one or more and work in coordination. The present invention does not exclude distributed processing, that is, the processor may be dispersed in different physical devices. The electronic device of the present invention is not limited to a single entity, but may also be the sum of multiple physical devices.

[0104] The memory stores a computer executable program, which is usually a machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least part of the steps in the method.

[0105] The memory includes a volatile memory, such as a random access memory unit (RAM) and / or a cache memory unit, and may also be a non-volatile memory, such as a read-only memory unit (ROM).

[0106] Optionally, in this embodiment, the electronic device further includes an I / O interface, which is used for the electronic device to exchange data with an external device. The I / O interface can represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0107] It should be understood that Figure 4 The electronic device shown is only an example of the present invention, and the electronic device of the present invention may also include elements or components not shown in the above examples. For example, some electronic devices also include display units such as display screens, and some electronic devices also include human-computer interaction elements such as buttons, keyboards, etc. As long as the electronic device can execute the computer-readable program in the memory to implement the method of the present invention or at least part of the steps of the method, it can be considered as an electronic device covered by the present invention.

[0108] Through the above description of the implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by combining software with necessary hardware. Figure 5 As shown, the technical solution according to the implementation mode of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes a number of commands to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the implementation mode of the present invention.

[0109] The software product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0110] The computer readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with a command execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0111] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0112] The computer-readable medium carries one or more programs. When the one or more programs are executed by a device, the computer-readable medium implements the data interaction method of the present disclosure.

[0113] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0114] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the exemplary embodiments described here can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiment of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several commands to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present invention.

[0115] It should be noted that the above detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present application belongs.

[0116] In the above detailed description, reference is made to the accompanying drawings, which form a part of this document. In the accompanying drawings, similar symbols typically identify similar components unless the context indicates otherwise. The illustrated embodiments described in the detailed description, drawings, and claims are not meant to be limiting. Other embodiments may be used, and other changes may be made, without departing from the spirit or scope of the subject matter presented herein.

[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A green bond index prediction method, characterized in that: include: Constructing influencing factor variables related to green bond market dynamics, wherein the influencing factor variables include at least one group of influencing factor variables; Collect influencing factor variables and green bond index logarithmic returns within a specified historical time period to establish a training data set, wherein the green bond index logarithmic returns are time series values ​​obtained after logarithmic transformation and difference processing in the time dimension; Based on the quantile adaptive group Lasso estimator, the established training data set is used for parameter estimation to construct a green bond index estimator; According to the parameter estimation values ​​obtained by parameter estimation, the constructed green bond index estimator is used to predict the dynamic changes of the green bond index.

2. The green bond index prediction method according to claim 1, characterized in that: The quantile-based adaptive group Lasso estimator uses the established training data set to perform parameter estimation to construct a green bond index estimator, including: Use the following expression to estimate the parameters and obtain the updated parameter estimates: in, are the parameter estimates, which is a vector where represents the intercept parameter estimate of the quantile adaptive group Lasso model obtained using historical data from time t0 to time t. It represents the estimated value of the regression vector of the quantile adaptive group Lasso model obtained using the historical data from time t0 to time t. Specifically, it is a multidimensional vector, including multiple parameters; α is the intercept parameter, β represents the regression coefficient vector, and λ is the penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, β i is the i-th parameter in the regression coefficient vector β, and the i-th parameter is related to the predicted variable p i,j Correspondingly, the value range of i is 1 to n, where n is the number of predictor variables, and the regression coefficient vector β has a group structure, that is, q is the number of groups, k is the group length index, is the regression coefficient vector β in g k The parameters in the group, and Among them, T k is the regression coefficient vector βg k The number of parameters in the group, is the gth in the regression coefficient vector β k The lth value in the group, Indicates g k The adaptive weights of the groups are used; the following groups of influencing factor variables are used to construct influencing factor variables related to the dynamics of the green bond market: energy source group, energy price group, stock market index group, monetary policy group and economic information group.

3. The green bond index prediction method according to claim 1, characterized in that: The quantile-based adaptive group Lasso estimator uses the established training data set to perform parameter estimation to construct a green bond index estimator, including: A penalty item parameter is selected based on a cross-validation method or a BIC value, and the value range of the penalty item parameter λ is greater than 0; g k Adaptive weights for parameters in a group The steps to determine the initial value of are: Step S210: using the quantile group Lasso model, obtain the following parameter estimation vector; in, are the initial parameter estimates, where It represents the initial intercept parameter estimate obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. It represents the estimated value of the initial regression vector obtained by using the quantile group Lasso model based on the historical data from time t0 to time t. a is the initial intercept parameter, b represents the initial regression coefficient vector, and γ is the initial penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, b i is the i-th parameter in the initial regression coefficient vector b. The i-th parameter is related to the predicted variable p i,j Correspondingly, the value range of i is 1 to n, n is the number of predictor variables, q is the number of groups, and k is the group length indicator. After grouping the initial regression coefficient vector b, g k The parameters in the group, and g represents the initial regression coefficient vector b k The hth parameter in the group, h=1,…,Z k , where Z k is the g of vector b k The number of parameters in the group, u k Yes k The penalty weight of the group is set Step S220: Further based on the obtained parameter estimation value, the adaptive weight is calculated using the following expression: in is the gth estimated value of the regression vector obtained in step S210 k The value of the group, is the gth regression vector estimate k The hth value in the group.

4. The green bond index prediction method according to claim 1, characterized in that: The parameter estimation values ​​obtained according to the parameter estimation are used to predict the dynamic change of the green bond index using the constructed green bond index estimator, including: Specifically, the following expression is used to predict the value of the logarithmic return of the green bond index: in, is the predicted value of the logarithmic return of the green bond index at time t+1, represents the intercept parameter estimate based on the historical data from time t0 to time t, represents the estimated value of the i-th regression parameter based on the historical data from time t0 to time t. The estimated value of the i-th regression parameter is related to the predicted variable p i,j Correspondingly, p i,t is the i-th predictor variable at time t, the value of i ranges from 1 to n, and n is the number of predictor variables.

5. The green bond index prediction method according to claim 1, characterized in that: Further including: The influencing factor variables also include at least one of the following economic variables: crude oil related parameters, natural gas related parameters, natural gas prices and crude oil prices in a designated country, and related indexes in a designated country.

6. The green bond index prediction method according to claim 4, characterized in that: include: The dynamic variables for predicting the future green bond index include predicting changes in the logarithmic returns of the green bond index, where the quantiles are selected to include values ​​including the median, and the values ​​are in the range of 0.1 to 1.

0.

7. The green bond index prediction method according to claim 4, characterized in that: include: The influencing factor variables and green bond index collected within a specified historical time period are used to establish a training data set and a test data set, and the training data set is used to perform parameter estimation to obtain parameter estimation values; The obtained parameter estimation values ​​are adopted and tested using the test data set to obtain prediction results, and the prediction results are compared with the true values ​​to further adjust and optimize the parameter estimation values.

8. A green bond index prediction device, characterized in that: It implements the green bond index prediction method according to any one of claims 1 to 7, and the green bond index prediction device comprises: A construction module, used to construct influencing factor variables related to the dynamics of the green bond market, wherein the influencing factor variables include at least one group of influencing factor variables; A collection and processing module, used to collect influencing factor variables and green bond indexes within a specified historical time period to establish a training data set, wherein the logarithmic return of the green bond index is a time series value obtained after logarithmic transformation and difference processing in the time dimension; A construction module is used to perform parameter estimation based on the quantile adaptive group Lasso estimator using the established training data set to construct a green bond index estimator; The prediction module predicts the dynamic change of the green bond index using the constructed green bond index estimator according to the parameter estimation values ​​obtained by parameter estimation.

9. The green bond index prediction device according to claim 8, characterized in that: include: Use the following expression to estimate the parameters and obtain the updated parameter estimates: in, are the parameter estimates, which is a vector where represents the intercept parameter estimate of the quantile adaptive group Lasso model obtained using historical data from time t0 to time t. It represents the estimated value of the regression vector of the quantile adaptive group Lasso model obtained using the historical data from time t0 to time t. Specifically, it is a multidimensional vector, including multiple parameters; α is the intercept parameter, β represents the regression coefficient vector, and λ is the penalty term parameter, which controls the intensity of the penalty on the regression coefficient. is the quantile loss function at time t, r j is the logarithmic return of the green bond index at time j, where j represents a certain time point or moment, specifically 1, 2, ..., t; p i,j is the i-th predictor variable at time j, β i is the i-th parameter in the regression coefficient vector β, and the i-th parameter is related to the predicted variable p i,j Correspondingly, the value range of i is 1 to n, n is the number of predictor variables, q is the number of groups, and k is the group length indicator. After grouping the regression coefficient vector β, g k The parameters in the group, and Among them, T k is the regression coefficient vector βg k The number of parameters in the group, Indicates g k Adaptive weights of groups; The following groups of influencing factor variables are used to construct influencing factor variables related to the dynamics of the green bond market: energy source group, energy price group, stock market index group, monetary policy group and economic information group.

10. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the green bond index prediction method described in any one of claims 1 to 7.

Images