창업이 고용변화에 미치는 영향에 관한 연구

Many policy makers as well as scholars believe that stimulating new firm formation is a promising way to achieve economic growth since existing firms are not creating a lot of jobs. So, developing countries, including South Korea, tend to implement government policies to boost new start-ups through industrial policy programs such as tax exemptions, employment subsidies, and streamlining start-up procedures. However, the empirical evidence concerning the effects of new firm formation on economic development is far from being entirely clear. We still do not have sufficient proof or knowledge about the ways in which new firm formation shapes employment changes and what time period it takes until the effects become visible in empirical data. This paper analyzes the short-term, medium-term and long-term effects of business start-ups on employment, and compares the degrees of the effects depending on the startup businesses' types of industries and levels of technologies. We calculate startup rates using the data from the Report on the Census on Basic Characteristics of Korean Enterprises by the Korean government (Korean Statistical Office), and estimate the relationship between the “start-up rates” and the rates of employment changes, using Almon polynomial lag procedure, which may reduce the multicollinearity problem using a lag distributed model. To decide the adequate length of lags, we use not only the Residual Mean Square, F-values, adjusted R2, and the Akaike Information Criterion(AIC) but also likelihood ratio tests. We do the homoskedasticity test and find that the data have heteroskedasticity. So, we use Huber-White Robust regression, which is known to correct the heteroskedasticity problem. Scholars usually use the “start-up rates” meaning one of the two approaches: the ecological approach (business stock approach) and the labor market approach. We use the latter approach which defines the start-up rate as the number of workers divided by the number of start-up businesses, as Garofoli (1994) suggests, from 1995 to 2006. In the process of analysis, we control for the population density of 2006, lagged dependent variable, and spatial autocorrelation. The population density of regions may have a high correlation with various characteristics of the regions including their development levels, labor quantities, and wage levels. The lagged independent variable is included because of the possible reversed causality that a high employment rate may cause a high start-up rate. The spatial autocorrelation variable is included because a change in the start-up rate of a region may interact with that of adjacent regions. We have observed that there are positive effects of business start-ups on employments for 1 to 10 years, which have three steps of the “direct effect,” the “replacement effect,” and the “inducement effect” like previous analyses such as Fritsch and Muller (2004). The effects of new business formation on employment changes over time in Korea show a wave (“S”) pattern, which is similar to the analysis on Germany (Fritsch and Muller, 2004), the United Kingdom (van Stel and Storey, 2004), Spain (Arauzo-Carod, Liviano-Solis and Marin-Bofarull, 2008), Portugal (Baptisa, Escaria, and Madrugo, 2008), South Korea's 7 major cities (Lee, 2009), and 21 OECD countries (Carree and Thurik, 2008). The direct effect from t to t-1 time periods increases the employment rate by 0.320%p, the replacement effect from t-2 to t-6 years decreases the employment rate by 0.396%p, and the inducement effect from t-7 to t-10 years increases the employment rate by 0.139%p. The direct effect is larger than the inducement effect in Korea, and the indirect effect period is longer than that of the previously studied countries. Taking the 3 effects altogether, start-ups increases the employment rate by 0.062%p, which means one unit of new start-up out of 1,000 employees increases the employment rate by 0.062%p for 10 years. The regression results have the adjusted R2 of 0.179. We have also shown that there are differences in the effects and the effecting periods among industries. Hospitality industries have a direct effect of 0.307%p, an indirect effect of -0.501%p, and an inducement effect of 0.224%p, and the overall effect is 0.029, which is the smallest in all industries due to the frequent entries and exits. Manufacturing industries show a direct effect of 1.276%p, an indirect effect of -2.677%p, and an inducement effect of 1.450%p, and the overall effect is 0.048%p, while the knowledge-based service industries have a direct effect of 3.170%p, an indirect effect of -3.229%p, and an inducement effect of 1.081%p, and its overall effect is 1.022%p, where there are the most pronounced ups and downs among the 3 categories. The analysis shows that the effect of start-ups on the employment rate is larger in the knowledge-based service sectors than in manufacturing ones, where the former has a stronger direct effect, while the latter has a stronger inducement effect. In addition, service sectors, such as hospitality industries, show a strong short-term job creation effect, meaning that start-ups in knowledge-based industries have the largest effect on the employment rate at least for 10 years, while start-ups in manufacturing industries may have a larger effect for longer than a 10 year period because of the lingering technology spillover effects. Therefore, we believe that balanced startup promotion strategies between types of industries are desirable for continuous job creation and economic growth. In addition, the paper analyzes that the effects of startups on the employment rate according to the level of technologies. The overall effects of start-ups of the high-, medium-, and low-technology sectors on the employment rate are -0.351%p, 0.982%p, and -0.208%p, respectively. The high technology manufacturing sectors have negative effects on the employment rate, which is contrary to our expectation, and which is probably because the time-horizon for the analysis is only 10 years and not long enough to catch the entire effect. Medium technology manufacturing start-ups, however, contribute a net increase of job creation due possibly to the shorter period of waiting time for mass production, while low technology start-ups do not. The analysis shows that not the number of startups but the quality of start-ups matters, meaning that high and innovative technology startups have an important effect on the market efficiency and overall job creation. The results also suggest that building the infrastructure for risk-taking for innovative start-ups can contribute economic growth through economic restructuring and hence the enhanced competitiveness. The paper analyzes the effect of start-ups for only 10 years, so a more general conclusion can be drawn for a longer time effect. In addition, we have not controlled for possible regional characteristics, which may have significant effects on the employment rate. Scholars could make a significant contribution to start-ups' job creation effects for further studies by considering demographic, industrial, environmental differences into analysis, using fixed effect or random effect models.