Keywords

Production Support, Public Programs, Entrepreneurship, Financing.

Introduction

Many countries promote entrepreneurial activity as a policy for social development and productivity growth (Storey, 2003). Multiple factors determine the success or failure of ventures sup-ported by public programs. Firstly, the internal factors of each venture: such as the previous experience of the founders (Cumming, 2006), the industrial sector to which the ventures belong (Watson et al., 1998), the existence of patents and the scope of the operation or financial resources owned by the company (Song et al., 2008). Secondly, and perhaps even more importantly for Economic Development Agencies (EDAs), is that they fail to distinguish between sound and bad projects to finance. The latter is a significant problem when promoting entrepreneurial activity and is related to the information asymmetries that exist between those who finance these projects and entrepreneurs, giving rise to an adverse selection problem (Cumming, 2006; Davidson & MacKinnon, 1992).

In Chile, CORFO has a wide variety of programs, which have different purposes and are oriented to different business development stages. For example, there are programs to support new entrepreneurs, where CORFO provides support by providing advice, mentoring and allocating competitive funds. Any entrepreneur can apply to these programs and may be subsequently selected or rejected depending on the venture's characteristics and whether it is considered a high-impact one in terms of its potential growth.

As proposed by Takalo et al. (2005) and Engel & Keilbach (2007), participants may be discouraged if they consider that they do not have the ability or potential to pass the selection process, self-selecting out of the capital market. Therefore, it can be inferred that entrepreneurs self-select when applying for this type of initiatives that promotes entrepreneurship (Autio & Rannikko, 2016).

As a result, it is possible for smaller ventures to self-select outside the capital market, re- sorting only to the capital from public funds such as CORFO, in contrast to those that can obtain external financing, who would favor private funds.

Literature Review and Hypothesis

There is extensive literature addressing high-growth ventures, including the impact of both public and private programs on their success, the difficulties in selecting the best ventures in terms of their high growth potential, and which attributes and characteristics these ventures tend to have (Friar & Meyer, 2003; Cumming, 2007; Goedhuys & Sleuwaegen, 2010; Cancino, 2014; Cancino et al., 2015; Cancino et al., 2018). The relevance of this resides in considering that the ventures which genuinely have a high impact on job creation and economic development are high-growth ventures (Wong et al., 2005; Rivas, 2014).

Goedhuys & Sleuwaegen (2010) analyzed 947 firms from 11 African countries and ten different sectors, using a quantum regression model, to determine the characteristics of high-growth ventures in that continent. Its results suggest that there would be substantial differences in the different sectors concerning to the growth level. The results show that there are indeed differences, where companies involved in product innovation, which have their means of transport and have websites, have higher growth rates.

Cancino (2014) studied the factors that may affect the internationalization of Chilean start-up firms. This paper analyzes a sample of 112 Small and Mid-Size Enterprises (SMEs) that regularly export. A logistic model was used to analyze the factors that affect the probability of a firm becoming international and thus a born global (BG). The results of this study indicate that Chilean firms are influenced by both national and international networks. In addition, the psychological distance from exporting countries would positively influence the probability of being a BG. At the same time, the technological level would not have an important influence on Chilean BG companies, which would be related to the fact that Chile relies mainly on the efficient exploitation of its natural resources.

Meanwhile, Friar & Meyer (2003) analyzed the factors that impact the development of ventures in the USA. To conduct this study, the authors used data from 90 business plans sent to a Boston competitor, separating the analysis into high-growth companies and microenterprises. This suggests that both exogenous and endogenous factors should be considered when fostering the development of high-growth ventures.

Self-Selection Problem at CORFO

Considering the above, it seems evident that there could be a problem of self-selection with the entrepreneurs who apply to the CORFO programs. Such self-selection would be inherent to this type of program, given that the same entrepreneurs decide whether to apply or not (Autio & Rannikko, 2016). The underlying problem with such self-selection lies in the fact that it is not necessarily the enterprises that ultimately benefit from it that have the most probability of high growth. Moreover, it seems reasonable to think that even more developed ventures, with greater potential and therefore, the greater probability of high growth, have fewer incentives to apply since they have the option of obtaining financial backing and support from a private agent. On the Contrary, smaller enterprises with lower development chances would have greater incentives to apply to programs such as CORFO's, as they would have fewer options to obtain private capital (Takalo et al., 2005; Engel & Keilbach, 2007; Colombo et al., 2007).

CORFO, in its different programs, applies filters and considers different criteria at the moment of selecting the benefited enterprises. The purpose of this is to support the ventures with the highest growth potential. However, the self-selection mentioned above makes this task difficult. Private investors also have the incentive to look for the best ventures since the profitability of their investment will depend on it. Consequently, the most reputable investors will tend to invest in better firms (Cumming & Dai, 2013). Nevertheless, as previously emphasized, those ventures considered less likely to grow, which would tend to be smaller, would be less likely to apply for private financing and, be more likely to apply for programs such as CORFO.

Hypothesis

The above suggests that programs such as CORFO's would have greater difficulty detecting high growth potential ventures. Thus, the most attractive ventures would prioritize other financing options. It could be a problem for the underlying objectives behind policies aimed at fostering innovation and country development.

In this respect, it seems reasonable to find important differences in the success or growth of emerging enterprises supported by CORFO, which could be explained by the simple fact that the company that applies for a certain program sees CORFO as a strategic element for its success, in which case it would have greater options to survive and grow, since the enterprise would have high potential. In contrast, there are new ventures that access CORFO mainly because they have no other option to finance themselves in the market.

Thus, enterprises that have received support from a CORFO program and are more established are expected to have a greater probability success. It would be related to the size of the company, the countries in which it operates, and its ability to differentiate its product or service. In turn, companies that have received capital funds, which indicates that they see CORFO as a strategic element and not their only option, would be firms that have greater growth potential and would have greater possibilities of surviving. Thus, the hypotheses are as follows:

H1: More established start-ups would have a greater probability of success in the market.

H2: New start-ups that see CORFO as a strategic option to grow would have a greater growth probability.

Data and Research Methodology

To perform an empirical analysis, we worked with data provided by CORFO's Public Entrepreneurship Database. The database considers 6001 project submissions to CORFO's entrepreneurship support programs between 2001 and 2018, and resources were allocated in the corresponding years. Since this research is aimed at studying the performance of enterprises that have recently joined CORFO, it was agreed to work with project data that were accepted between 2012 and 2017. On the other hand, due to the required adjustments to perform the estimations, finally, the study database considered a sample of 317 firms. In a later stage, to conduct the second estimation, work was carried out with a subset of the previous sample, which considered a total of 231 observations.

In view of this data, the research method considered two models. The first was a logistic regression model, which sought to test the first hypothesis, according to which variables associated with greater business consolidation would be associated with a greater probability of success. The logistical model that seeks to represent the probability of success in a venture can be represented as follows:

In turn, to test the second hypothesis, a multinomial logistic regression model was implemented, which is commonly used for multiple result qualitative response problems (McFadden, 1976). In this case, the model contemplates three possible scenarios and can be represented through the Softmax function as follows:

In this second model, we sought to study the characteristics that affect firms' evolution and development in each of the three possible scenarios. A summary of the key variables is presented in Table 1.

First Model: Logistic Regression

The first logistical regression model seeks to explain the success of ventures benefited by a CORFO program. The dependent variable "Success" is a binary variable, and it is measured through the sales level and the company's ability to survive. Therefore, the Success variable takes the value 1 if the venture effectively reached sales in 2017 and in turn managed to survive. If any of the two previous conditions (or both) is not met, the variable takes the value 0.

Second Model: Multinomial Logistic Regression

The second model, which refers to a multinomial logistic regression, seeks to explain the growth possibility of projects that were benefited by a program belonging to CORFO. The dependent variable "Growth" is a categorical variable and is measured through the level of sales in 2017 as compared to the level of sales in 2016. Consequently, the variable Growth takes the value 1 (decrease) if the company had fewer sales in 2017 compared to 2016, the value 2 (steady) if the sales level remained the same in 2017 compared to 2016, and 3 (growth) if the sales level increased in 2017 compared to 2016. The independent variables for the two models are presented in Table 1.

Table 1 Key Variables
Variable	Type	Description
N° Countries	Continuous variable ranging from 0 to 12	Total number of countries where the firm sells
Work Experience	Continuous variable ranging from 1 to 35	Years of work experience
Diversity	Dummy variable takes a value of 0 for no and 1 for yes.	Are there at least two members of different gender?
Previous Entrepreneurship	Dummy variable takes a value of 0 for no and 1 for yes.	Has at least one member made a previous entrepreneurship?
Internationalize	Dummy variable takes a value of 0 for no and 1 for yes.	Has the enterprise been internationalized?
National Capital	Dummy variable takes a value of 0 for no and 1 for yes.	Has the company received private national funds?
Foreign capital	Dummy variable takes a value of 0 for no and 1 for yes.	Has the company received private international funds?
Workers	Categorical variable. 1 if you do not have employees; 2 if you have between 1 and 3 employees; 3 if you have 4 or more employees.	size of the company
Program	Categorical variable. 1 if you belong to “Others”; 2 for PRAE; 3 for SSAF Innovation; 4 for SSAF Social.	program that has received support
Sector	Categorical variable. 1 for Advertising; 2 for Biomedicine & biotech; 3 for Commerce; 4 for Education; 5 for Financial & business services; 6 for Fisheries & aquaculture; 7 for Food; 8 for Fruit, wine and port; 9 for Health & pharmaceutical; 10 for Manufacture; 11 for Mining & metalworking; 12 for Others; 13 for Social; 14 for TCIE; 15 for TIC.	Industrial sector
Product type	Categorical variable. 1 for Hardware; 2 for Physical product; 3 for Service; 4 for Software.	Product category
Patent	Categorical variable. 1 for It has patent; 2 for It has not been done; 3 for It is in process; 4 for It is not patentable.	Does the product have a patent?

Results

To understand which variables may be most relevant at the time of identifying the enterprises with the greatest potential, a logistic regression model was first estimated. The results are shown in Table 2. Subsequently, a multinomial logistic regression model was estimated to identify the factors related to the likelihood of growing, decreasing or maintaining the sales level from one period to another. The results of this second model can be seen in Table 3.

From the logistic regression model presented in Table 2, five variables are statistically significant at 1%: workers, patent, and number of countries, internationalization, and foreign capital. Thus, the results suggest that companies which have more workers would be related to a greater probability of success in the business once the threshold of four workers is passed. In other words, larger enterprises that benefited would have greater chances of success. In turn, failing to patent the product or service is negatively associated with the company's probability of success. The same relationship occurs for the non-patentable product; however, it has a lower impact on the probability of success and is significant at 5%. On the other hand, having a greater number of countries increases the probability of success of these ventures, which could be due to greater diversification. Finally, raising foreign capital also increases the likelihood of a successful venture, which appears to be consistent with prior statements.

Table 2 Logistic Regression Model
Model	Coeff. Est	95% CI	Z	P-value
Success Workers (base 1=0)
=2, 1 to 3	0.093	(-0.018 - 0.203)	1.638	0.101
=3, 4 or more	0.176***	(0.066 - 0.286)	3.133	0.002
Program (base 1=Others)
=2, PRAE	0.015	(-0.079 - 0.109)	0.319	0.75
=3, SSAF Innovation	-0.045	(-0.145 - 0.055)	-0.883	0.377
=4, SSAF Social	-0.051	(-0.387 - 0.285)	-0.297	0.766
Sector (base 1=Advertising)
=2, Biomedicine & biotech	0.033	(-0.255 - 0.321)	0.223	0.823
=3, Commerce	0.005	(-0.246 - 0.256)	0.038	0.969
=4, Education	0.002	(-0.266 - 0.269)	0.014	0.989
=5, Financial & business serv	0.145	(-0.077 - 0.368)	1.278	0.201
=6, Fisheries & aquaculture	0.052	(-0.277 - 0.381)	0.309	0.758
=7, Food	0.080	(-0.162 - 0.322)	0.648	0.517
=8, Fruit, wine and hort	-0.310	(-0.700 - 0.079)	-1.56	0.119
=9, Health & pharmaceutical	-0.019	(-0.280 - 0.242)	-0.141	0.888
=10, Manufacture	0.059	(-0.209 - 0.327)	0.433	0.665
=11, Mining & metalworking	0.025	(-0.251 - 0.301)	0.181	0.857
=12, Others	0.114	(-0.109 - 0.336)	0.998	0.318
=13, Social	0.042	(-0.247 - 0.331)	0.286	0.775
=14, TCIE	0.165	(-0.057 - 0.388)	1.458	0.145
=15, TIC	0.043	(-0.189 - 0.275)	0.365	0.715
Product type (base 1=Hardware)
=2, Physical product	0.053	(-0.077 - 0.183)	0.794	0.427
=3, Service	0.050	(-0.083 - 0.183)	0.739	0.46
=4, Software	0.060	(-0.069 - 0.188)	0.913	0.361
Previous entrepreneurship	-0.051	(-0.155 - 0.054)	-0.951	0.342
Work experience	-0.003	(-0.007 - 0.002)	-1.107	0.268
Patent (base 1 = It has patent)
=2, It has not been done	-0.135***	(-0.212 - -0.058)	-3.424	0.001
=3, It is in process	-0.140*	(-0.299 - 0.018)	-1.732	0.083
=4, It is not patentable	-0.095**	(-0.168 - -0.022)	-2.539	0.011
Diversity	0.023	(-0.048 - 0.094)	0.628	0.53
N°Countries	0.179***	(0.115 - 0.242)	5.528	0
Internationalize	-0.096***	(-0.167 - -0.026)	-2.674	0.007
National Capital	0.065	(-0.017 - 0.147)	1.559	0.119
Foreign capital	0.201***	(0.074 - 0.329)	3.094	0.002
Observations	317

Concerning the multinomial logistic regression model presented in Table 3, the probability of decreasing sales (decrease), there are significant differences between sectors, where enterprises in education, fishing, and manufacturing would be less likely to decrease their sales over time. At the same time, it can be observed that firms belonging to the RESP program are more likely to decrease their sales as well as firms that decide to internationalize. In the case of the probability of maintaining a stable level of sales (steady), it is possible to see that work experience is related to more stable ventures, although only at 10% of significance. It is an interesting finding that could be associated with a wider contact network that would maintain greater certainty in the business but would not imply greater growth. Finally, in the probability of increasing the sales level, it may be observed that firms that received foreign private capital would have greater possibilities to growing over time, a sign that they are potential high-growth firms.

Table 3 Logistic Regression Model
	Decrease		Steady		Growth
Model	Coeff	P-value	Coeff	P-value	Coeff	P-value
Growth Workers (base 1=0)
=2, 1 to 3	0.012	0.879	-0.119	0.233	0.108	0.262
=3, 4 or more	-0.062	0.424	-0.096	0.378	0.158	0.129
Program (base 1=Others)
=2, PRAE	0.266**	0.022	-0.215*	0.077	-0.052	0.628
=3, SSAF Innovation	0.035	0.618	-0.136	0.14	0.102	0.235
=4, SSAF Social	0.010	0.926	-0.015	0.951	0.005	0.984
Sector (base 1=Advertising)
=2, Biomedicine & biotech	0.068	0.691	-0.348	0.106	0.28	0.23
=3, Commerce	0.026	0.858	0.097	0.59	-0.123	0.376
=4, Education	-0.187**	0.044	-0.084	0.655	0.27	0.135
=5, Financial & business serv	-0.017	0.871	-0.213	0.164	0.23	0.119
=6, Fisheries & aquaculture	-0.187**	0.044	-0.087	0.881	0.273	0.636
=7, Food	-0.002	0.991	-0.134	0.489	0.136	0.44
=8, Fruit, wine and hort	0.092	0.737	-0.599***	0	0.508*	0.079
=9, Health & pharmaceutical	-0.096	0.438	-0.277	0.146	0.373**	0.045
=10, Manufacture	-0.187**	0.044	0.212	0.334	-0.025	0.906
=11, Mining & metalworking	0.125	0.46	0.089	0.639	-0.214*	0.079
=12, Others	0.028	0.807	-0.13	0.394	0.102	0.461
=13, Social	0.065	0.699	-0.187	0.371	0.122	0.566
=14, TCIE	-0.056	0.64	-0.216	0.224	0.272	0.106
=15, TIC	-0.159	0.107	-0.142	0.38	0.301*	0.056
Product type (base 1=Hardware)
= 2, Physical product	-0.176	0.101	-0.087	0.531	0.263**	0.03
= 3, Service	-0.037	0.778	-0.178	0.205	0.215*	0.063
= 4, Software	-0.041	0.74	-0.02	0.884	0.061	0.558
Previous Entrepreneurship	-0.053	0.435	-0.092	0.327	0.145	0.129
Work Experience	0.001	0.895	0.011*	0.057	-0.012**	0.039
Patent (base 1=It has patent)
=2, It has not been done	-0.102	0.38	0.151	0.33	-0.049	0.747
=, It is in process	0.069	0.716	-0.027	0.888	-0.042	0.845
=4, It is not patentable	-0.023	0.848	0.111	0.427	-0.089	0.506
Diversity	-0.015	0.77	0.041	0.582	-0.026	0.709
N°Countries	0.016	0.252	0.002	0.91	-0.019	0.284
Internationalize	0.107**	0.026	-0.035	0.618	-0.072	0.254
National Capital	0.038	0.506	0.064	0.453	-0.102	0.201
Foreign Capital	-0.043	0.469	-0.125	0.135	0.168**	0.031
Observations	231
Z-score indicates significance level as follows *** p<0.01, ** p<0.05, * p<0.1

Discussion

Some firms that have been approached by CORFO show signs of having more chances of success and potential growth. This result may imply that, within the set of ventures that apply to CORFO programs, certain types of ventures possess better features for selection. If analyzed considering the existing information asymmetries, particularly the adverse selection problem at the time of granting benefits and self-selection raised in this paper, enterprises of greater potential choose CORFO as a strategic option and others for which it is the only alternative. Precisely, it seems that there are conditions associated with the size, the exclusivity of the product or service, and the capital raised that would be a positive sign regarding the firm's potential. Likewise, the fact that a firm has received private capital could be a good sign of the project's degree of innovation and, consequently, of its potential success in the market.

The essential condition that could explain a company's growth seems to be the private capital received. As mentioned above, foreign private capital would be a sign of the quality of a venture in terms of its ability to function in the market and its ability to grow. It implies that a private investor's decision to invest in a project would most likely be a good indication of the project's potential and differentiation capacity. As discussed above, the results show that it seems easier to find promising ventures for the market than for a government agent, given that the same applicants self-select. Some findings are noteworthy. According to the findings, internationalization would be related to value destruction towards the firm. Although initially, these results might seem unintuitive, Alvarez (2004) points out that there is a great difference in the percentage of successful exporters if we compare small and mid-size companies (SMEs) with large companies. Of all exporting SMEs, only 7% are successful, while in large companies, this figure is 40%. This is due to the fact that small firms would have greater limitations related to heterogeneity in access to market information and managerial skills, economies of scale, among others.

Conclusion

The present study attempted to analyze the results of ventures benefited by a CORFO program in Chile, evaluating whether certain characteristics would give signs of better performance. The goal is to identify attributes or qualities that will allow mitigating the adverse selection and self-selection problem that would exist in these public programs, distinguishing between projects that strategically choose CORFO (firms with high growth potential) and projects that choose it because they have no other option. This research could be very significant to understand better the asymmetric information problems in programs such as those of CORFO at the time of selecting projects and the difficulties that underlie the efforts made to effectively select the highest quality ventures. The findings of this study can therefore provide valuable information for the decisions made by organizations such as CORFO in terms of which characteristics and criteria to most consider when selecting projects from which the best possible results are expected. The above, not only a way to improve the use of the resources available by these institutions, but also as an important pillar regarding the country's economic development through the success of these innovative initiatives. Within this research's scope, when a venture has received support from a CORFO program and has also been supported by a private agent, the results of such ventures are significantly better than those of companies that have been supported by CORFO alone. On the other hand, the ventures that show signs of greater stability in the market could show greater success chances. This could be useful to generate better criteria and selection mechanisms within CORFO, facilitating the identification of high-growth enterprises that can become large companies that generate jobs and facilitate economic growth post pandemic.

References

1. Alvarez, R. (2004). Sources of exliort success in small-and medium-sized enterlirises: The imliact of liublic lirograms. International Business Review, 13(3), 383-400.
2. Cancino, C.A. (2014). Raliid Internationalization of SMEs: Evidence from" Born Global" Firms in Chile. Innovar: Revista de ciencias administrativas y sociales, 141-151.
3. Cancino, C.A., Bonilla, C.A., &amli; Vergara, M. (2015). The imliact of government suliliort lirograms for the develoliment of businesses in Chile. Management Decision, 53(8), 1736-1754.
4. Cancino, C.A., La liaz, A.I., Ramalirasad, A., &amli; Syn, T. (2018). Technological innovation for sustainable growth: An ontological liersliective. Journal of Cleaner liroduction, 179, 31-41.
5. Colombo, M.G., Grilli, L., &amli; Verga, C. (2007). High‐tech start‐uli access to liublic funds and venture caliital: Evidence from Italy. International Review of Alililied Economics, 21(3), 381-402.
6. Cumming, D. (2006). Adverse selection and caliital structure: Evidence from venture caliital. Entrelireneurshili Theory and liractice, 30(2), 155-183.
7. Cumming, D. (2007). Government liolicy towards entrelireneurial finance: Innovation investment funds. Journal of Business Venturing, 22(2), 193-235.
8. Cumming, D., &amli; Dai, N. (2013). Why do entrelireneurs switch lead venture caliitalists?. Entrelireneurshili Theory and liractice, 37(5), 999-1017.
9. Davidson, R., &amli; MacKinnon, J.G. (1992). Regression-based methods for using control variates in Monte Carlo exlieriments. Journal of Econometrics, 54(1-3), 203-222.
10. Engel, D., &amli; Keilbach, M. (2007). Firm-level imlilications of early stage venture caliital investment-An emliirical investigation. Journal of Emliirical Finance, 14(2), 150-167.
11. Friar, J.H., &amli; Meyer, M.H. (2003). Entrelireneurshili and start-ulis in the Boston region: Factors differentiating high-growth ventures from micro-ventures. Small Business Economics, 21(2), 145-152.
12. Goedhuys, M., &amli; Sleuwaegen, L. (2010). High-growth entrelireneurial firms in Africa: A quantile regression aliliroach. Small Business Economics, 34(1), 31-51.
13. McFadden, D.L. (1976). Quantal choice analysis: A survey. Annals of Economic and Social Measurement, 5(4) 363-390.
14. Rivas, G. (2014). El fomento al emlirendimiento dinámico en liaíses del Cono Sur de América Latina: Análisis de exlieriencias y lecciones de liolítica. Banco Interamericano de Desarrollo, IDB-Dli-363.
15. Song, M., liodoynitsyna, K., Van Der Bij, H., &amli; Halman, J.I. (2008). Success factors in new ventures: A meta‐analysis. Journal of liroduct Innovation Management, 25(1), 7-27.
16. Storey, D.J. (2003). Entrelireneurshili, small and medium sized enterlirises and liublic liolicies. In Handbook of Entrelireneurshili Research. Sliringer, Boston, MA, 473-511.
17. Takalo, T., Tanayama, T., &amli; Toivanen, O. (2005). Selection or self-rejection? Alililications into a voluntary treatment lirogram: The case of R&amli;D subsidies. Helsinki Center of Economic Research (76).
18. Watson, K., Hogarth‐Scott, S., &amli; Wilson, N. (1998). Small business start‐ulis: success factors and suliliort imlilications. International Journal of Entrelireneurial Behavior &amli; Research, 4(3), 217-238.
19. Wong, li.K., Ho, Y.li., &amli; Autio, E. (2005). Entrelireneurshili, innovation and economic growth: Evidence from GEM data. Small Business Economics, 24(3), 335-350.