Rethinking the government as innovator: Evidence from Asian firms

Abstract

Can government ownership of a firm spur innovation? While a robust body of empirical and theoretical work suggests that government ownership suppresses innovation, the increasing global competitiveness of government-owned firms coupled with the transition of these firms’ competitive advantage from low-cost labor-focused to innovation-focused suggests we reconsider the relationship between government ownership and innovation. This paper develops and tests a theory that a minority level of government ownership has a positive influence on firm-level innovation, by substituting for a better-developed financial market and alleviating financial constraints. By combining the incentives and monitoring benefits of private ownership with the more “patient” shareholders of and protection from failure by the state, minority government ownership can provide an environment for firms to pursue otherwise risky innovative projects. The testing of this theory makes use of a large sample of Asian firms, and the results with respect to matched minority government-owned and privately held firms show that minority government ownership increases the likelihood of innovation among financially constrained firms; moreover, it especially increases the likelihood of process innovation (as opposed to product innovation), which is traditionally difficult to finance. The study’s results—which stand to benefit firms and governments worldwide—remain robust to alternative mechanisms and controls, as well as to a variety of sensitivity tests.

Appendix

Variables

Dependent variables

The World Bank Enterprise Survey asked responders whether in the past 3 years their firm had undertaken any of the following seven innovation activities: (1) developed a major new product line, (2) upgraded an existing product line, (3) introduced a new technology that substantially changed the way the main product line is produced, (4) brought in-house a major production activity that had been outsourced, (5) outsourced a production activity that had been in-house, (6) obtained a licensing agreement, and/or (7) formed a joint venture with a foreign partner. INNOV is the variable that is the main measure of firm-level innovation used for this analysis, is a binary measure equal to 1 if any of these activities had indeed been pursued during the 3 years in question. Analyses examining process innovation consider innovation as a binary measure equal to 1 if the firm introduced a new technology that substantially changed the way the main product line is produced. And analyses examining product innovation consider innovation as a binary measure equal to 1 if a firm developed a major new product line or upgraded an existing product line.

These non-traditional measures present several advantages relative to those typically used to assess developed markets, such as patents and R&D expenditure. In the EE context, where underdeveloped property rights make patenting less valuable, and where technological change tends to result from adapting foreign technologies rather than creating technology that is new to the entire world (Acemoglu, Aghion, & Zilibotti, 2006), patents and R&D expenditure are arguably less accurate measures of innovation. Thus, for EEs where firms are more likely to engage in imitation and adoption rather than patenting, the measures of innovation outlined in the preceding paragraph are more representative. Furthermore, the measures used in this paper have more policy accuracy in the sense that they are modeled after the measures of innovation established by the Oslo Manual for use in developing countries (OECD, 2005). These measures are consider innovation as “new to the firm”, not necessarily “new to the world.”

The measures used here also avoid two potential problems associated with relying on R&D expenditure as a measure of innovation. The first is that R&D expenditure is a noisy measure of innovation, as not all R&D expenditure leads to innovation, whereas the measures used here express only innovative activities that are actually pursued. The second potential problem is that formal R&D measures are typically biased against smaller firms. This limitation is even more problematic in the EE context, where small and medium-sized firms are significant drivers of growth (see e.g., Archibugi & Sirilli, 2001; Michie, 1998). However, the multi-dimensional nature of the innovation measure used here, along with the distribution in firm size helps mitigate this bias.

Independent variables

In this study, the main construct for minority government ownership is a binary measure equal to 1 if the government owns a share in the firm that is at least 10% but less than 50%. The moderating variable that we are interested in, financial constraints, is measured as an indicator equal to 1 for firms that reported experiencing significant obstacles to conducting business because of financial constraints. In particular, respondents are asked to indicate on a five-point scale the severity of the problems (if any) that lack of access to financing presents, with higher values indicating that a lack of access to financing presented larger problems for the operation and growth of the firm.¹⁰ It is possible to capture the existence of barriers to accessing external financing by collapsing this variable into an indicator variable equal to 1 for firms that reported experiencing moderate, major, or very severe obstacles as a result of limited financing, and into an indicator variable equal to zero for firms reporting no or minor obstacles to operations and growth as a result of limited financing. In the sample of 11 countries, approximately 40% of the firms reported experiencing moderate, major, or severe obstacles as a result of financing constraints.

Before the main analysis, I also conduct a side external validity test to substantiate the self-reported measure of constraints. The basic idea is that countries where a higher proportion of firms report financial constraints should also be less well-developed in terms of their financial markets. Thus, if the self-reported measures are accurate, the proportion of firms with constraints should be positively related to financial development. Thus, for each country, I calculate the proportion of firms that report financing constraints.

I then adopt several widely-used measures of country-level financial market development from the World Bank Financial Development and Structure Dataset, defining financial development in terms of financial market size, efficiency, and liquidity (Beck, Demirgüç-Kunt, & Levine, 2000). To proxy for elements of financial sector size, I measure liquid liabilities per GDP (llgdp), bank deposits, per GDP (dbagdp), and stock market capitalization (stk mkt cap). Liquid liabilities is a broad measure of financial resources, including all banks, bank-like and non-bank financial institutions. It is measured as currency plus demand and interest-bearing liabilities of banks and other financial intermediaries divided by GDP. The second measure of banking sector development, bank deposits per GDP, represents the bank deposits theoretically available to the banking sector for lending. Stock market capitalization represents a measure of size of a financial market. Altogether, a higher value of these measures indicates a better-endowed financial market. Thus, these values should be negatively correlated with the proportion of firms in an economy reporting financial constraints.

Unlike the size measures, measures of efficiency are constructed from raw bank-level data from the BankScope database and averaged (unweighted) across all banks of a country for a given year. Three measures of efficiency are used, including net interest margin (net int), overhead costs (overhead) and banking sector concentration (conc). Net interest margin equals the accounting value of a bank’s net interest revenue as a share of its total earning assets. Overhead cost equals the accounting value of a bank’s overhead costs as a share of its total assets. Higher levels of these measures indicate lower levels of banking efficiency, as banks incur higher costs and there is a larger spread between lending and deposit interest rates. The final measure of efficiency is banking sector concentration, measured as the ratio of the three largest banks’ assets to total banking sector assets. A high concentration ratio suggests a less competitive banking market structure. Thus, these measures should be positively correlated with the proportion of firms in an economy reporting financial constraints.

Finally, to provide a measure of the activity/liquidity of the capital market, I use stock market total value traded to GDP (stk mkt val) and the stock market turnover ratio (stk mkt turn). Stock market total value traded to GDP equals total shares traded on the stock market exchange divided by GDP, and it indicates the activity of the stock market trading volume as a share of national input and should reflect the degree of liquidity that stock markets provide to the economy. The stock market turnover ratio, in contrast, measures the activity or liquidity of a stock market relative to its size. A small but active stock market will have a low turnover ratio. It is measured as the value of total shares traded divided by market capitalization.¹¹

Other important control variables include company age and size, foreign ownership, exporting, capacity utilization, and internal cash flow. Age is the number of years since the firm started operations. It is unclear how a firm’s age may affect its innovation since older firms may have developed change-resistant routines yet at the same time may also have accumulated the innovation-promoting capabilities and knowledge. To facilitate interpretation of regression coefficients, size is measured as an indicator equal to 1 for those firms with greater than the median number of employees (although using the natural logarithm of employees produced similar results in terms of direction and significance). The standard reasoning for including size as a variable potentially correlated with innovation is that large companies have more financial and other resources that may be put toward innovation; furthermore, larger companies can typically benefit from economies of scale in innovative activity.

Foreign ownership is measured as the percentage of firm equity that is foreign-owned. Foreign ownership is typically associated with the increased likelihood of innovation, as firms are exposed to the increased knowledge pool afforded by foreign parents. Also, activity in export markets may expose the firm to additional technology and/or force it to become more competitive. In this study this factor is controlled for with the percentage of sales that come from exported products. Whether capacity utilization should be positively or negatively correlated with innovation ex-ante is unclear. Capacity utilization is the percentage of a firm’s output relative to maximum possible output. If firms are at capacity, they may need to innovate in order to facilitate future sales, suggesting that capacity utilization may positively influence innovative activity. However, firms may also be more interested in extending current capacity (if demand is high) than in pursuing innovation (Ayyagari, Demirgüç-Kunt, & Maksimovic, 2011), suggesting a negative relationship. Finally, internal cash flow is measured as the percentage of investment financed by internal funds. As the World Bank Enterprise Survey does not address innovation financing specifically, this study takes all internal funding earmarked for investment to be positively correlated with investment in innovation. Also included are industry, country, and year dummies.

In order to illustrate differences between innovation activities, Table 6 shows the summary statistics of the above main variables for the minority government-owned firms as well as for the privately owned firms within the full sample of Asian firms. T-tests of the differences in means show that minority government-owned firms are on average older, larger, and less likely to innovate. This is consistent with the prior literature that shows that government firms tend to be older, concentrated largely in industries with larger capital requirements, and, when under political influence, vulnerable to the possibility of inflated payrolls (as discussed earlier). Table 6 also shows that both minority government-owned and privately held firms include large outliers in terms of firm size, as the mean and median for firm size are quite different. Table 7 provides the correlation matrix for the main explanatory variables, showing no severe multi-collinearity among independent and control variables.

Table 6 Summary statistics of key variables

Table 7 Correlation matrix of dependent and explanatory variables for Asian firms in the World Bank Enterprise Survey