Empirical Economics

, Volume 52, Issue 4, pp 1367–1378 | Cite as

Do credit guarantees for small and medium enterprises mitigate the business cycle? Evidence from Korea

Article

Abstract

This study empirically examines the impact of credit guarantees for small and medium enterprises (SMEs) on the business cycle in Korea, using quarterly time series data over the period 2002–2012. Credit guarantees are denoted by the total amount of credit guarantees provided by three authorities in Korea, and the business cycle consists of two volatilities, that is, real GDP per capita and industrial product index. We used Toda and Yamamoto Granger causality test and autoregressive distributed lag (ARDL) bounds test approach to examine the existence of cointegration and to find out the short-run and long-run relationship between credit guarantees and business cycle. The ARDL bounds testing approach has an advantage that it can be used when I (0) and I (1) variables are mixed. Based on the ARDL bounds test approach, we find that credit guarantees for SMEs mitigate business cycle fluctuations in the short run, but find no significant effects in the long run.

Keywords

Credit guarantees Business cycle SMEs ARDL bounds test 

JEL Classification

E32 G38 

1 Introduction

Active small and medium enterprises’ (SMEs) activities are important for reducing monopolistic systems, protecting consumer sovereignty, improving income inequality, and providing opportunities for increased output growth and employment. In addition, SMEs comprise 99.9 % of enterprises and account for 86.9 % of total employment in Korea.1 Accordingly, the activities of the SMEs in Korea are central to local development as well as economic growth.

Nonetheless, SMEs lack funds for investment because of the asymmetry of information in financial markets, which becomes an obstacle in promoting a country’s economic growth. Hence, most governments actively support SMEs with policy funds for preserving the function of financial institutions and removing the uncertainty and risk of the market. It is widely recognized that credit guarantees are a typical policy fund for SMEs in many countries. That is, credit guarantees can correct both market failure and coordination failure and become the driving force for the development of SMEs.

However, some literature, such as Beck et al. (2008) and Cowling (2010), argued that credit guarantees distort the resource allocation because they supply funds to SMEs that have a lower level of feasibility, inducing a possible reduction in self-sustenance and hindering the reconstruction of SMEs. In other words, guarantee scheme can increase moral hazard among borrowers by reducing the default risk they otherwise would incur (i.e., by providing part of the collateral required to obtain the loan). This can lead to more “strategic defaults” from borrowers, because part of the collateral does not belong to the borrower, which makes a higher incentive to default (OECD 2009).

Hence, the existing literature has examined the effectiveness of policies that fund supply using credit guarantees through data on various countries (for instance, Boocock and Shariff 2005; Cowling 2010; General Audit Office 2007; Kang and Heshimati 2008; Larrain and Quiroz 2006; Lelarge et al. 2008; Oh et al. 2009; Riding et al. 2007; Santarelli and Vivarelli 2002). The studies mainly focused on microeconomic analyses and showed that credit guarantees enhance the corporate performance of innovative SMEs or technology-oriented SMEs. However, the results for common SMEs depended on data, sample selections, and time periods.

Several existing macroeconomic analyses have suggested that active SMEs’ activities are related to the business cycle. For example, Gertler and Gilchrest (1994) demonstrated that both sales and storage decrease to a greater extent in SMEs than in large enterprises in response to a contractionary monetary policy shock.2 In addition, Sharpe (1994) showed that employment growth at more highly leveraged firms is more sensitive to demand and financial market conditions over the business cycle. Hence, the policy fund for SMEs may be related with the business cycle. Nevertheless, few studies have analyzed the direct relationship between the policy fund for SMEs and the business cycle.3 The results possibly provide some implications about the countercyclicality of credit guarantee for SMEs.

The main motivation of this study is to estimate the impacts of credit guarantees for SMEs on the business cycle in Korea, by using quarterly data over the period 2002–2012. The existing literature (for instance, Lim et al. 2011; Rhee 2011) used Korean data focused on the relationship between credit guarantees and economic growth. However, it is necessary to examine the relationship between credit guarantees and business cycle, since the business cycle is closely related to the economic growth (for instance, Aizenman and Marion 1999; Chong and Gradstein 2009; Ramey and Ramey 1995).

To denote the policy fund using credit guarantees, we use the total amount of credit guarantees provided by three authorities. Furthermore, the business cycle is computed by using the volatilities with real GDP per capita and industrial product index. We also use M2 monetary aggregate, considering for the impacts of monetary policy or external factors.4 The autoregressive distributed lag (ARDL) bounds testing approach to cointegration is used for empirical analysis in order to find out the short-run and long-run relationship between credit guarantees and business cycle. The ARDL bounds testing approach has an advantage that it can be used when I (0) and I (1) variables are mixed, because it does not require all variables to be I (1) or I (0). In addition, the advantage of the ARDL bounds testing approach method is that it can be used even in the cases of small sample sizes (Narayan 2005; Halicioglu 2007), as in this paper. Therefore, the analysis of this study provides an implication as to whether policy fund using credit guarantees mitigates the business cycle in Korea.

The remainder of this paper is organized as follows. Section 2 describes the empirical methodology, including data descriptions, empirical equations, and estimation methods. Section 3 presents a series of empirical results. Section 4 concludes the paper.

2 Data and empirical model

This study examines the long-run and short-run impacts of credit guarantees for SMEs on the business cycle in Korea using quarterly data over the period 2002–2012. The total amount of credit guarantees (henceforth TACG) is the sum of credit guarantees provided by Korea Credit Guarantee Fund, Korea Technology Finance Corporation, and Local Credit Guarantee Foundation,5 and these data are obtained from the related authorities.6 The business cycle consists of two volatilities, real GDP per capita (henceforth VOL1) and industrial product index (henceforth VOL2), and both data are drawn from the Bank of Korea.7 Data on the M2 monetary aggregate (henceforth M2), which are from the economic statistics of the Bank of Korea, are used to control for the impacts of monetary policy.

In particular, we use the HP (Hodrick–Prescott) filter method to remove trends caused by cyclical fluctuations in the business cycle data. Furthermore, we convert the data to seasonally adjusted variables using Census Bureau X-12 ARIMA (autoregressive moving average) method. Table 1 contains summary statistics for above three variables. In addition, Fig. 1 illustrates how the variables changed over time. Both VOL1 and VOL2 showed a very similar pattern, and TACG was sharply increased in 2008–2009 to come up with an effective counterplan for the global financial crisis.
Table 1

Summary statistics for variables

 

TACG

VOL1

VOL2

M2

Mean

9.48

−0.01

−0.30

14.02

Median

9.45

0.24

0.15

13.99

Maximum

9.85

2.57

5.75

14.42

Minimum

9.16

−4.68

−16.73

13.57

Standard deviation

0.14

1.48

4.20

0.27

Skewness

0.30

−1.03

−1.80

0.03

Kurtosis

2.89

4.80

7.65

1.52

Jarque–Bera [Prob.]

0.67 [0.71]

13.70 [0.00]

63.54 [0.00]

4.01 [0.13]

Observations

44

44

44

44

(1) TACG, VOL1, VOL2, and M2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” “industrial product index volatility,” and “M2 monetary aggregates,” respectively. (2) All variables are natural logarithms and seasonally adjusted using Census Bureau X-12 ARIMA

Fig. 1

Plots of variables. Note VOL1, VOL2, TACG, and M2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” “industrial product index volatility,” and “M2 monetary aggregates,” respectively

First of all, we conduct unit root tests to verify the stationarity of the time series data. That is, we employ the traditional methods of the augmented Dickey–Fuller (ADF) test and Phillips–Perron (PP) test for this purpose. To determine the optimal lags in the unit root tests, we consider all plausible cases of “constant term,” “constant term \(+\) trend,” and “none” using the Schwarz information criteria (henceforth SIC). The results in Table 2 suggest that TACG is integrated of order one, namely I (1), and that both VOL1 and VOL2 are I (0) in all cases.

The ARDL bounds testing approach is used as the empirical method, because I (0) and I (1) variables are mixed. The ARDL bounds testing approach, developed by Pesaran and Shin (1999) and Pesaran et al. (2001), estimates as a single equation based on the unrestricted error correction model (henceforth UECM) to test for the existence of cointegration. In other words, the ARDL bounds testing approach can be used to test for the existence of cointegration, regardless of whether the explanatory variables are stationary or nonstationary. In addition, the advantage of the ARDL bounds testing approach is that it can be used even in the cases of small sample sizes. In comparison with a standard cointegration test,8 this method has an advantage as regards classification of dependent and independent variables. That is, the classification of variables in a standard cointegration test is unclear because the method is based on a VAR (vector autoregression) analysis. By contrast, the classification of variables in the ARDL bounds testing approach based on a single equation analysis is clear.

When we use an equation with two variables, the structure of UECM for the ARDL bounds testing approach takes the following form:
$$\begin{aligned} \Delta \ln Y_t= & {} b_0 + \sum _{a=1}^n b_{1a} \Delta \ln Y_{t-a} + \sum _{a=1}^n b_{2a} \Delta \ln X_{t-a} \nonumber \\&+\,b_3 \ln Y_{t-1} + b_4 \ln X_{t-1} +e_t . \end{aligned}$$
(1)
Here, \(b_0\) is a constant, \(b_{1a}\) and \(b_{2a}\) are the short-run coefficients, \(b_3\) and \(b_4\) are the long-run coefficients, and \(e_t\) is a white noise error term. The ARDL bounds test is based on the Wald test (F-statistic), and the null hypothesis is \(b_3 = b_4 =0\), which indicates the nonexistence of a long-run relationship. If the calculated F-statistic exceeds the upper critical bound, the series are cointegrated. If it is below the lower critical bound, there is no cointegration. If the calculated F-statistic is between the upper critical bound and the lower critical bound, the existence of cointegration is inconclusive.9
The ARDL bounds testing approach uses a \(\left( {n+1} \right) ^{k}\) formula to estimate the number of regressions. The n indicates the maximum number of lags utilized, and k denotes the total number of variables. The lag length is selected using the minimum values of the Akaike information criteria (henceforth AIC) and the SIC. In addition, as long as a cointegrating relationship has been detected between two variables, an ARDL model can be established to determine the long-run and short-run relationships as follows:
Table 2

Results of unit root tests

Variable

Type

ADF test

PP test

Level

1st differences

Level

1st differences

TACG

Intercept

−2.15 [0.23]

−7.94*** [0.00]

−2.05 [0.26]

−8.06*** [0.00]

Intercept \(+\) trend

−2.52 [0.32]

−7.84*** [0.00]

−2.52 [0.32]

−7.96*** [0.00]

None

0.21 [0.74]

−8.03*** [0.00]

0.31 [0.77]

−8.13*** [0.00]

VOL1

Intercept

−3.99*** [0.00]

−4.96*** [0.00]

−2.59 [0.10]

−5.74*** [0.00]

Intercept \(+\) Trend

−3.93** [0.02]

−4.90*** [0.00]

−2.56 [0.30]

−5.59*** [0.00]

None

−4.03*** [0.00]

−5.01*** [0.00]

−2.63*** [0.00]

−5.83*** [0.00]

VOL2

Intercept

−4.34*** [0.00]

−5.81*** [0.00]

−2.83* [0.06]

−5.64*** [0.00]

Intercept \(+\) Trend

−4.31*** [0.00]

−5.74*** [0.00]

−2.79 [0.21]

−5.54*** [0.00]

None

−4.39*** [0.00]

−5.89*** [0.00]

−2.85*** [0.00]

−5.82*** [0.00]

M2

Intercept

−0.36 [0.91]

−3.17** [0.03]

−0.82 [0.80]

−3.15** [0.03]

Intercept \(+\) Trend

−1.83 [0.67]

−3.13 [0.11]

−1.24 [0.89]

−3.11 [0.12]

None

2.55 [0.99]

−1.84* [0.06]

7.87 [1.00]

−1.81* [0.07]

(1) TACG, VOL1, VOL2, and M2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” “industrial product index volatility,” and “M2 monetary aggregates,” respectively. (2) p values are provided in brackets. (3) (***), (**), (*) significant at 1, 5, 10 % levels, respectively

Long-run relationship:
$$\begin{aligned} {\ln }Y_t =a_0 + \sum _{i=1}^k a_{1i} \ln Y_{t-i} + \sum _{j=1}^l a_{2j} \ln X_{t-j} +e_t , \end{aligned}$$
(2)
Short-run relationship:
$$\begin{aligned} \Delta \ln Y_t =b_0 +b_1 \left( {ECT} \right) _{t-1} + \sum _{p=1}^n \ln Y_{t-p} + \sum _{q=1}^m b_{2q} \Delta \ln X_{t-q} +e_t . \end{aligned}$$
(3)
Here, kln, and m are the lag lengths of single variables based on AIC and SIC, and \(ECT_{t-1}\) is an error correction term resulting from the estimated long-run equilibrium. Using the results of Eqs. (2) and (3), long-run coefficients (\(\beta _0\) and \(\beta _1\)) in Eq. (4) and short-run coefficients (\(\alpha _0\) and \(\alpha _1\)) in Eq. (5) can be calculated.
$$\begin{aligned} \ln Y_t= & {} \beta _0 +\beta _1 \ln X_t +e_t \end{aligned}$$
(4)
$$\begin{aligned} \Delta {\ln }Y_t= & {} \alpha _0 +\alpha _1 \Delta \ln X_t +e_t \end{aligned}$$
(5)
Using the parameter estimates of Eqs. (2) and (3), the long-run coefficients (\(\beta _0\) and \(\beta _1\)) and the short-run coefficients (\(\alpha _0\) and \(\alpha _1\)) are calculated as a form of nonlinear functions as follows.
$$\begin{aligned} \beta _0= & {} \frac{a_0 }{\left( {1- {\sum }_{i=1}^n a_{1i} } \right) }, \quad \beta _1 = \frac{ {\sum }_{j=0}^m a_{2j} }{\left( {1- {\sum }_{i=1}^n a_{1i} } \right) } \end{aligned}$$
(6)
$$\begin{aligned} \alpha _0= & {} \frac{b_0 }{\left( {1- {\sum }_{p=1}^n b_{1p} } \right) }, \quad \alpha _1 = \frac{ {\sum }_{q=0}^m b_{2q} }{\left( {1- {\sum }_{p=1}^n b_{1p} } \right) } \end{aligned}$$
(7)
Hence, we construct a UECM to examine the existence of cointegration between the two variables. In this process, we use the Toda–Yamamoto Granger causality test for establishing the dependent variable \(Y_{t}\) and the independent variable \(X_{t}\). If cointegration exists, we compute the long-run and short-run coefficients from the results of the ARDL model. In addition, we consider time dummy variables for 2008 and 2009 to accommodate the influence of the 2008 global financial crisis in our model.

3 Empirical results

First of all, the Toda–Yamamoto Granger causality test is conducted to identify the causality between credit guarantees and the business cycle. Toda and Yamamoto (1995) approach is based on a standard VAR model for the levels of the variable instead of for the first-order differences in the variables as in the Granger causality tests. This approach can minimize the risks resulting from the possibility of wrong detection of the cointegration levels of the series. A VAR model with two variables can be formed as follows:
$$\begin{aligned} Y_t= & {} \alpha + \sum _{i=1}^k \beta _{i } Y_{t-i} + \sum _{j=k+1}^{k+d} \beta _j Y_{t-j} + \sum _{i=1}^k \gamma _i X_{t-i} \nonumber \\&+\,\sum _{j=k+1}^{k+d} \gamma _j X_{t-j} + \epsilon _{1t}, \end{aligned}$$
(8)
$$\begin{aligned} X_t= & {} \omega + \sum _{i=1}^k \delta _{i } X_{t-i} + \sum _{j=k+1}^{k+d} \delta _j X_{t-j} + \sum _{i=1}^k \theta _i Y_{t-i} \nonumber \\&+\,\sum _{j=k+1}^{k+d} \theta _j X_{t-j} + \epsilon _{2t}. \end{aligned}$$
(9)
Here, k is the number of lags, and d represents the maximum cointegration level of the variables entered into the model. The basic idea of this approach is to increase the number of lags in the VAR model in accordance with the maximum cointegrating relationship of the series for a \(\chi ^{2}\) distribution. The null hypothesis for Eq. (8) is \(H_0 : \gamma _i =0\) and that for Eq. (9) is \(H_0 : \theta _i =0\). If the null hypothesis is rejected by the Wald test in Eq. (8), X is the cause of Y; however, if the null hypothesis is rejected by the Wald test in Eq. (9), Y is the cause of X.
This study conducts a Toda–Yamamoto Granger causality test in the VAR model using X as the TACG and Y as the business cycle. The results of this test are shown in Table 3. The results indicate that the null hypotheses are rejected at the conventional significance level in all cases. Hence, we can identify an interrelationship (or a bidirectional causal relationship) between TACG and the business cycle.
Table 3

Toda–Yamamoto Granger causality tests

Null hypothesis

Time lag \(\left( {k+d}\right) \)

\(\upchi ^{2}\)

p value

TACG \(\nRightarrow \) VOL1

5

23.40

0.00

VOL1 \(\nRightarrow \) TACG

5

18.78

0.00

TACG \(\nRightarrow \) VOL2

5

13.50

0.02

VOL2 \(\nRightarrow \) TACG

5

12.64

0.03

TACG, VOL1, and VOL2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” and “industrial product index volatility,” respectively

Accordingly, we compose two UECM models. That is, each model uses TACG and the business cycle as dependent variables. In addition, the model that uses VOL1 is named Model 1 and the one that uses VOL2 is named Model 2. We choose the proper lag length using AIC and SIC to implement the ARDL bounds test and report the results in Table  4.
Table 4

Optimal number of lags

Dependent variable

Model 1 (VOL1 is used)

Model 2 (VOL2 is used)

AIC

SIC

AIC

SIC

TACG

−2.42 (1)

−2.05 (1)

−2.43 (1)

−2.06 (1)

Business cycle

2.73 (2)

3.14 (1)

4.67 (2)

5.18 (2)

(1) TACG, VOL1, and VOL2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” and “industrial product index volatility,” respectively. (2) The optimal number of lags is provided in parentheses

In Table  4, both the AIC and SIC recommend an optimal lag length of 1 in Models 1 and 2 when TACG is used as the dependent variable. When the business cycle is used as the dependent variable, the optimal number of lags is 2 (AIC) and 1 (SIC) in Model 1, and 2 (AIC) and 2 (SIC) in Model 2. Using the optimal number of lags, we conduct the Wald test for the differential variables of the dependent as well as independent variables based on the UECM to identify the existence of cointegration. The results are summarized in Table  5.
Table 5

Results of ARDL bounds test

Dependent variable

Model 1 (VOL1 is used)

Model 2 (VOL2 is used)

AIC

SIC

AIC

SIC

TACG

1.21

1.21

1.97

1.97

Business cycle

6.36***

5.34***

7.97***

7.97***

ARDL bounds test, critical value (Pesaran et al. 2001)

Significance level

Critical value of upper bound

Critical value of lower bound

1 %

5.61

4.29

5 %

4.35

3.23

10 %

3.77

2.72

(1) TACG, VOL1, and VOL2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” and “industrial product index volatility,” respectively. (2) (***) is significant at 1 % significance level

The result of the ARDL bounds testing approach indicates that the cointegration between TACG and the business cycle exists in the cases when the business cycle is used as the dependent variable. Models 1 and 2 of the ARDL bounds testing results based on AIC are larger than the critical value of the upper bound provided by Pesaran et al. (2001) at the 1 % significance level. However, the results of other cases10 are smaller than the critical value of the lower bound at the 10 % significance level, and hence a cointegrating relationship between TACG and the business cycle does not exist. Therefore, TACG is suggested to be the cause of the business cycle in the long run.

In the cointegration cases, we set up long-run and short-run models and use AIC and SIC to determine the optimal number of lags for each model. Based on the estimations of the long-run and short-run models, it is possible to induce long-run and short-run coefficients as in Eqs. (6) and (7), which implies long-run and short-run impacts of TACG on the business cycle, respectively. The estimation results of the long-run and short-run coefficients are given in Table 6.
Table 6

Empirical results of the long-run and short-run coefficient

Variable

Model 1 (VOL1 is used)

Model 2 (VOL2 is used)

Long-run coefficient

   Constant term

16.93 (0.73)

13.88 (0.22)

   TACG

−4.21 (−1.30)

−3.17 (−0.35)

   M2

1.63 (0.10)

1.17 (0.22)

Short-run coefficient

   Constant term

−0.73 (−1.25)

−1.01 (−0.91)

   \(\Delta \) TACG

−18.12*** (−3.38)

−30.07*** (−3.07)

   \(\Delta \) M2

40.33 (1.37)

61.08 (1.14)

(1) TACG, VOL1, VOL2, and M2 denote “total amount of credit guarantees,” “real GDP per capita volatility,” “industrial product index volatility,” and “M2 monetary aggregates,” respectively. (2) t values are provided in parentheses. (3) (***), (**), (*) significant at 1, 5, 10 % levels, respectively

The results in Table 6 show that all long-run coefficients of TACG are statistically insignificant, whereas short-run coefficients are negative and statistically significant at the 1 % significance level (both Model 1 and Model 2). This implies that the business cycle is not affected by the changes in TACG in the long run. However, an increase in TACG reduces the fluctuations of the business cycle in the short run. The result is plausible if we believe that the business cycle is more closely related to the short run rather than to the long run. In other words, the results suggest the existence of countercyclicality of credit guarantees in the short run. The impacts of credit guarantees for SMEs on economic growth and/or economic fluctuation in the long run are ambiguous or negligible: However, those help the business confrontation and delay the exit of marginal SMEs including a possible reduction in self-sustenance in the short run.

In addition, Fig. 2 illustrates the results of CUSUM (or cumulative sum control chart) test based on the cumulative sum of recursive residuals, which show the stability of parameters. It is found that both plots of CUSUM against critical bound at the 5 % significance level verify the structural stability of ARDL model.
Fig. 2

Plots of cumulative sum of recursive residuals. a Cumulative sum of recursive residuals in Model 1. b Cumulative sum of recursive residuals in Model 2. Note The model that uses VOL1 (“real GDP per capita volatility”) is named Model 1 and the one that uses VOL2 (“industrial product index volatility”) is named Model 2

Hence, the results suggest that the credit guarantees for SMEs contribute to reducing the short-run fluctuation of the business cycles in Korea, which is consistent with the existing arguments of Noh (2009) and Sharpe (1994). In other words, our results imply that credit guarantees not only provide funding opportunities for SMEs with poor credit and mortgage abilities but also reduce the volatility of the business cycle in the short run, which mitigates the uncertainty of the economy.

4 Concluding remarks

This study empirically examines the impact of credit guarantees for SMEs on the business cycle in Korea, using quarterly data over the period 2002–2012. The total amount of credit guarantees (TACG) is the sum of credit guarantees provided by Korea Credit Guarantee Fund, Korea Technology Finance Corporation, and Local Credit Guarantee Foundation. The data are obtained from the related authorities. The business cycle consists of two volatilities, real GDP per capita and industrial product index, and both are calculated using the HP (Hodrick–Prescott) filter method. In addition, M2 monetary aggregate is controlled to accommodate the impacts of monetary policy.

We use the ARDL bounds testing approach as an empirical method, because both I (0) and I (1) variables are mixed [i.e., TACG is I (1) and business cycle variables are I (0)]. Next, we conduct the Toda–Yamamoto Granger causality test and identify that bidirectional relationships exist. The result of the ARDL bounds test suggests that the cointegrating relationship between TACG and the business cycle exists in the cases of models where the business cycle is used as the dependent variable. In cases of cointegration, we are able to determine the short-run and long-run coefficients.

The regression results suggest that the short-run impact of TACG on the business cycle is negative and statistically significant at the conventional significance level without model specifications. However, the long-run coefficients are statistically insignificant without model specifications. Based on the findings, we conclude that credit guarantees for SMEs possibly provide the short-run countercyclicality. That is, the impacts of credit guarantees for SMEs on economic fluctuation in the long run are negligible: However, those help the business confrontation and delay the exit of marginal SMEs in the short run.

Footnotes

  1. 1.

    These data are available from the Small and Medium Business Administration in Korea (http://www.smba.go.kr).

  2. 2.

    Similar results are found in Chari et al. (2007) by expanding the time periods.

  3. 3.

    The relationship between financial development and output volatility is provided in Easterly et al. (2000) and Kose et al. (2006). For example, Easterly et al. (2000) showed that greater credit or a deeper financial system is significantly associated with less volatility in all specifications. The consumption and production smoothing possibilities provided by the existence of a deep financial system might reduce growth volatility.

  4. 4.

    See Gali (2008) for more information about the relationship between monetary policy and business cycle.

  5. 5.

    Korea Credit Guarantee Fund is organized to extend credit guarantees for the liabilities of promising SMEs which lack tangible collateral. Korea Technology Finance Corporation is a specialized institution in providing full-scale supports to SMEs and venture businesses with competitive technology, innovation, and other knowledge-based business contents at all growth stages. Local Credit Guarantee Foundation was created to revitalize the regional economy by providing financial support to SMEs, as well as to small retailers that lack security.

  6. 6.

    Because data on the Local Credit Guarantee Foundation are available from 2002, we use quarterly data over the period 2002–2012.

  7. 7.

    These may be downloaded from the economic statistics system of the Bank of Korea (http://ecos.bok.or.kr).

  8. 8.

    For example, the Johansen (1988) efficient approach has been widely used to test for the existence of cointegrating relationships.

  9. 9.

    The critical bounds are taken from Pesaran et al. (2001).

  10. 10.

    In other words, the cases when TACG is used as the dependent variable.

Notes

Acknowledgments

We have benefited from comments and suggestions on the editor of this journal and anonymous referees. Any remaining errors or ambiguities, of course, are ours.

References

  1. Aizenman J, Marion N (1999) Volatility and investment: interpreting evidence from developing countries. Economica 66:157–179CrossRefGoogle Scholar
  2. Beck T, Klapper LF, Mendoza JC (2008) The typology of partial credit guarantee funds around the world. Policy research working paper No. 4771, World BankGoogle Scholar
  3. Boocock G, Shariff MNM (2005) Measuring the effectiveness of credit guarantee schemes: evidence from Malaysia. Int Small Bus J 23:427–454CrossRefGoogle Scholar
  4. Chari VV, Christiano L, Kehoe P (2007) The Gertler–Gilchrist evidence on small and large firm sales. Working paper (http://faculty.wcas.northwestern.edu/~lchrist/research/cck/notesongertlergilchristdecember2006.pdf)
  5. Chong A, Gradstein M (2009) Volatility and growth. J Econ Growth 14:1–25CrossRefGoogle Scholar
  6. Cowling M (2010) Economic evaluation of the small firms loan guarantee (SFLG) scheme. Report commissioned by the UK Department of Business Innovation and SkillsGoogle Scholar
  7. Easterly W, Islam R, Stiglitz JE (2000) Shaken and stirred: explaining growth volatility. In: Pleskovic B, Stern N (ed) Annual World Bank Conference on DevelopmentGoogle Scholar
  8. Economic Statistics System of the Bank of Korea (http://ecos.bok.or.kr/)
  9. Gali J (2008) Monetary policy, inflation, and the business cycle: an introduction to the new Keynesian framework. Princeton University Press, PrincetonGoogle Scholar
  10. General Audit Office (2007) Small business administration: additional measures needed to assess 7 (a) loan program’s performance. GAO Report 07-769, USAGoogle Scholar
  11. Gertler M, Gilchrest S (1994) Monetary policy, business cycles, and the behavior of small manufacturing firms. Q J Econ 109:309–340CrossRefGoogle Scholar
  12. Halicioglu F (2007) Residential electricity demand dynamics in Turkey. Energy Econ 29:199–210CrossRefGoogle Scholar
  13. Johansen D (1988) Statistical analysis of cointegrated vectors. J Econ Dyn Control 12:231–254Google Scholar
  14. Kang JW, Heshimati A (2008) Effect of credit guarantee policy on survival and performance of SMEs in Republic of Korea. Small Bus Econ 31:445–462CrossRefGoogle Scholar
  15. Kose MA, Prasad ES, Terrones ME (2006) How do trade and financial integration affect the relationship between growth and volatility? J Int Econ 69:176–202CrossRefGoogle Scholar
  16. Larrain C, Quiroz J (2006) Estudio para el fondo de garantía de pequeños empresarios. Mimeo, Banco del EstadoGoogle Scholar
  17. Lelarge C, Sraer D, Thesmar D (2008) Entrepreneurship and creditconstraints: evidence from a French loan guarantee program. In: International differences in entrepreneurship, NBER books, University of Chicago PressGoogle Scholar
  18. Lim H-J, Kwon E-J, Yoo S-h (2011) Does public credit guarantee contribute to economic growth? Evidence from Korea. Written in Korean. Asia Pac J Small Bus 33:5–15Google Scholar
  19. Narayan PK (2005) The saving and investment nexus for China: evidence from cointegration tests. Appl Econ 37:1979–1990CrossRefGoogle Scholar
  20. Noh YH (2009) Countercyclicality of credit guarantee and its implications: evidence from the Korea credit guarantee fund. Written in Korean. Korean Small Bus Econ 31:185–206Google Scholar
  21. OECD (2009) Facilitating access to finance. Discussion paper on credit guarantee schemes. ParisGoogle Scholar
  22. Oh I, Lee JD, Heshimati A, Choi GG (2009) Evaluation of credit guarantee policy using propensity score matching. Small Bus Econ 33:335–351CrossRefGoogle Scholar
  23. Pesaran MH, Shin Y (1999) An autoregressive distributed-lag modelling approach to cointegration analysis. In: Econometrics and economic theory in the 20th century. Cambridge University Press, CambridgeGoogle Scholar
  24. Pesaran MH, Shin Y, Smith RJ (2001) Bound testing approaches to the analysis of level relationship. J Appl Econom 16:289–326CrossRefGoogle Scholar
  25. Ramey G, Ramey V (1995) Cross country evidence on the link between volatility and growth. Am Econ Rev 85:1138–1151Google Scholar
  26. Rhee CO (2011) Business cycle and SME’s government-supported financing. Written in Korean. Asia Pac J Small Bus 33:17–32Google Scholar
  27. Riding A, Madill J, Haines G (2007) Incrementality of SME loan guarantees. Small Bus Econ 29:47–61CrossRefGoogle Scholar
  28. Santarelli E, Vivarelli M (2002) Is subsidizing entry an optimal policy? Ind Corp Change 11:39–52CrossRefGoogle Scholar
  29. Sharpe S (1994) Financial market imperfections, firm leverage, and the cyclicality of employment. Am Econ Rev 84:1060–1074Google Scholar
  30. The Small and Medium Business Administration, Korea Report. (http://www.smba.go.kr)
  31. Toda HY, Yamamoto T (1995) Statistical inference in vector autoregressions with possibly integrated processes. J Econom 66:225–250CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2016

Authors and Affiliations

  1. 1.IBK Economic Research InstituteSeoulRepublic of Korea
  2. 2.Chungnam Techno ParkCheonanRepublic of Korea
  3. 3.Department of EconomicsHannam UniversityDaejeonRepublic of Korea

Personalised recommendations