Scholars have observed psychological factors as significant influencers of venture investment decision-making at early times. Duxbury et al. (1996) described a personality profile of angel investors with an internal locus of control, high needs for achievement, dominance, and autonomy. In addition, Mason and Stark (2004) mentioned psychological differences of angel investors compared with other types of investors, angel investors have stronger motivation of involvement to acquire “satisfaction and enjoyment from playing a role in the entrepreneurial process,” to be altruistic at times, and to give greater emphasis to subjective factors and gut instinct. Later studies on angel investors focused on their decision-making process, pointing out the use of heuristics, gut feel, and motivational clues as important machanisms (Maxwell et al., 2011; Harrison et al., 2015; Huang and Pearce, 2015; Cardon et al., 2017). In crowdfunding, herd effect and regulatory focuses were found to be influential (Ciuchta et al., 2016; Shen et al., 2020). Meanwhile, few studies have discussed the decision-making of an accelerator as a new organizational format in the entrepreneurial ecosystem. Yang et al. (2020) found that gender role congruity also exists in investment decisions in social impact accelerators. Meanwhile, congruity has been studied not only in terms of gender-role dyads but also in terms of cognitive modes (causation/effecuation) in entrepreneurial investment (Murnieks et al., 2011; Balachandra et al., 2017). Furthermore, studies have suggested that heuristics, gut feel, control motivation, and dispositional affect also matter in entrepreneurial investment decisions (Chan and Park, 2013, 2015; Drover et al., 2014; Huang, 2018). However, few studies have dealt with how a single psychological factor influences the preferences of investors for ventures with different features, especially at the organizational level.

Temporal orientation has been found to be an influential psychological mechanism, as “time is essentially in the eye of the beholder (Hall, 1984) and varies across people” (Lin et al., 2019, 3115). Several temporal frameworks elaborate on how people perceive and regard time (Trompenaars and Hampden-Turner, 1998; Souitaris and Maestro, 2010), among which the lens of long-term and short-term orientation has attracted wide interest (Laverty, 1996). It offers an anchor for people to consider and balance the focus on the past, present, and future (Karniol and Ross, 1996; Bearden et al., 2006) to form a dominant temporal logic for decisions and actions (Lumpkin and Brigham, 2011).

Based on the achievements of past studies concerning LTO, it is not only a macro culture dimension at the national level (Hofstede, 2001) but also a multidimensional construct of personal psychology (Lumpkin and Brigham, 2011). In the latest literature, LTO is defined as the “tendency to prioritize the long-range implications and impact of decisions and actions that come to fruition after an extended time period” (Lumpkin et al., 2010). Concern for the future is intuitively and actually a key attribute of LTO (Brigham et al., 2013). However, a more accurate and complete understanding of LTO contains a holistic view of time with extended time horizon valuing both the past and the future instead of only caring for the effects of action in the here and now or the short term (Bearden et al., 2006; Lumpkin and Brigham, 2011). Futurity, continuity, and perseverance are the three components of LTO (Bearden, 2006; Lumpkin and Brigham, 2011).

Decision process is hardly possible to be entirely impartial. The orientations and perceptions of decision-makers are mirrored in their decisions while sifting through and reconciling incomplete ambiguous chaotic information (Hambrick et al., 1996). Thus, in intertemporal choices, LTO seems to be a significant force (Lumpkin and Brigham, 2011) widely discussed in strategic decision-making and family businesses literature. Long-term oriented top executives are willing to pursue interests in a farsighted and inclusive way (Miller and Le Breton-Miller, 2006), operate strategic control rather than financial control (Zahra et al., 2004), accelerate the development and deployment of new products (Yadav et al., 2007; Nadkarni and Chen, 2014), and maintain relationships with stakeholders (Flammer and Bansal, 2017). However, they take fewer strategic risks (Gentry et al., 2016). Meanwhile, LTO also improves the comprehensiveness, speed, and creativity of the strategic decision-making process (Lin et al., 2019) as well as innovative and entrepreneurial actions (Hofstede, 1991; Flammer and Bansal, 2017).

Venture capitals invest in new ventures in order to exit through acquisitions or IPOs with considerable returns. However, outcomes vary in terms of whether they can achieve this goal, how long it takes, and how much they will be paid back. Majority of venture investments fail to generate positive returns, behind which stands the fact that new ventures are always accompanied by unpredictable uncertainties of varying types and temporal distributions (Kollmann and Kuckertz, 2010; Huang, 2018). VCs consider these attributes of ventures and balance different distributions of uncertainties, costs, and gains over time (Souder and Bromiley, 2012) to form their investment strategies. With limited amount of money to allocate in the duration of the funds, they reach an investment decision according to criteria under the implicit guidance of a temporal orientation ranging from short to long (Harrison et al., 2015). Thus, the author proposes that VCs with high level of LTO will demonstrate several features.

As mentioned above, the author performed content analysis to measure the LTO of VCs, and first collected a list of VC organizations from IT Juzi (Su and Lichun, 2020), a website widely used by investors in China. Similar to VentureXpert, it aims to provide the most complete data on Chinese entrepreneurship and investments. It offers three streams of data: (1) founding information on new ventures, including founding dates, founding teams, and descriptions of business ventures; (2) VC profiles, including the amount of their managed money, employees, and other fundamental information; (3) funding information, including investor(s) and invested venture as well as investment date and amount of money. Specifically, the database contains founding and investing data for more than 200,000 ventures, 120,000 investment events, and 10,000 investment organizations, greatly exceeding data volume of other databases (e.g., another frequently used database named PE Daily records only approximately 23,000 investment events). The author matched these VCs with their WeChat official accounts if available. Nowadays, as digital technologies have reformed communication channels, organizations have been using these advancements to build their images and compete for the attention of the audience on the internet. WeChat has become the dominant communication tool and an important media in China. It has also become the most important platform for information dissemination. Organizations can create official accounts on WeChat. Their accounts serve as billboards for organizations to communicate with their audience, release latest news, and promote their ideas. For example, one VC posted an article to promote its investment logic:

“… In terms of early-stage projects, our advantage is not the amount of capital but the fact that we can do a lot with a little. We look at projects with a keen eye. Early-stage projects are indeed risky, but the high growth of a particular project will cover the loss of other projects…Entrepreneurship and investment are subject to various risks and challenges. Challenges are good, and risks are not necessarily bad. I am partial to investing in higher-risk angel rounds, especially those with resources…”

The author collected articles published on the WeChat official account of VCs by year using frequencies of LTO keywords in these articles to indicate the level of LTO of VCs. The details are introduced in the latter part of this article.

The matched sample included 606 VCs with 1,473 VC-year observations of LTO from 2012 to 2019¹. The author collected other data concerning general information and investment records of VCs from IT Juzi that were supplemented by other databases (such as PE daily and Tianyancha) and official websites of VCs.

The majority of past studies used questionnaires to measure LTO, implying the same premise that the time orientation of top executives represents this aspect of the organization (Hambrick and Mason, 1984). However, the studies seem to evade the question of how to weigh the influence of each executive on organizations under varying levels of LTO (Lin et al., 2019). Thus, it is meaningful to search for direct measurement of such psychological constructs at the organizational level.

Prior research has used firm disclosures (such as annual reports and letters for shareholders) to indicate their visions and intentions (Gerstner et al., 2013; Flammer and Bansal, 2017). With the growing acceptance of cognitive-linguistic theory, content analysis has got wider use in entrepreneurship literature (Lee and Huang, 2018; Pan et al., 2020). According to the Whorf–Sapir hypothesis (Whorf, 1940; Sapir, 1944), thoughts of individuals and organizations (attention, cognitive categories, priorities, etc.) are reflected in the words and vocabularies they use. In this vein, the author attempted to use the method of content analysis to measure the LTO of VCs with frequencies of LTO words used.

Flammer and Bansal (2017) constructed a simple vocabulary of LTO with only four pairs of words: “long run,” “long-run,” “long term,” and “long-term.” Nonetheless, the other dimensions of LTO, except for futurity, were ignored. Then, the author followed the method of Pan et al. (2020) to create a dictionary (vocabulary) with a complete indication of LTO constructs and tested its validity.

Given that a Chinese sample was used, the author applied the simplified Chinese dictionary of LIWC as the basic dictionary, because it includes all commonly used words and is widely accepted in academic research (Parhankangas and Renko, 2017)². First, a PhD student and the author, equipped with theoretical bases of LTO, independently selected keywords representing the construction of LTO from the basic dictionary using the scales from Bearden et al. (2006) as reference. These keywords constituted the original dictionary of LTO. In the second step, five PhD holders/PhD students independently voted on each keyword in the original dictionary, gauging whether the word reflected the meaning of LTO. They were also asked to make their judgments in accordance with the same scale. In the last step, the author calculated the content-validity ratio (CVR) for each keyword. The formula is:

where n denotes the votes each word obtained from the judges ranging from 0 to 5, and N is the number of judges (5 in this study). The CVR ranges from −1 to 1.0, and the author only retained keywords with a CVR greater than or equal to 0.8 (which means it received at least four out of five votes from the judges). Finally, the author used 44 keywords to construct the final LTO dictionary.

To test the validity of the LTO dictionary, the author used a random subsample containing 149 articles from the full sample collected. The author and the PhD student who joined the first step to create the original dictionary rated these articles on a five-point Likert scale, with 1 indicating “shows no LTO of VC at all” and 5 indicating “heavily shows the LTO of VC.” Meanwhile, the author measured the LTO indexes of the articles using the LIWC dictionary approach, dividing the total number of LTO keywords by the total number of words of each article. The author calculated ICC = 0.83, which supported the good validity of the linguistic measure of LTO.

Tables 1, 2 show the description and correlations of the variables. LTO did not exhibit any high correlations with the dependent variables and control variables, whereas some of the controls were correlated with one another. Investment experience was highly correlated with Industrial horizon and Network of VCs, which was understandable, as the more investments a VC made, the higher possibilities for it to invest in different industries and form ties with other VCs. Network also had a high correlation with Industrial horizon, as a larger network could offer more information and knowledge for the VC to explore opportunities in diverse industries. High correlations (above 0.5) appeared in other dyads among Number of investors, Investment experience, Investment accomplishments, Industrial horizon, and Network as well. The remaining controls, namely, Number of branches, Number of foreign branches, Age, and Amount of money managed, were not highly correlated with each other. The author applied OLS regression to test the hypotheses as expressed by Equation (1) below:

where i represents each VC in the sample, and t represents each calendar year. Thus, DV_i,t, LTO_i,t, and Control_i,t represent the dependent variables, LTO value, and control variables of each observation, respectively. p_i and y_t are the fixed effects of places and calendar years. As the sample is of the VC-year level, the author clustered the errors to the VC level in the estimation.

The results of the regression models are displayed in Tables 3, 4. The results did not support hypothesis 1, as the LTO coefficient in the model of H1 was not significant [β = −0.016, t = −0.34, p = 0.732, and 95% CI = (−0.108, 0.076)]. As for the models of H2, the LTO showed significant negative effects on the industrial popularity of the portfolio [β = −0.008, t = −2.72, p = 0.007, and 95% CI = (−0.014, 0.002)]. Thus, hypothesis 2 was supported, that is, VCs with higher level of LTO tend to have less “popular” portfolio. Hypothesis 3 was also supported by a significantly positive coefficient of LTO [β = 0.056, t = 2.54, p = 0.011, and 95% CI = (0.013, 0.099)], which indicated that VCs with higher level of LTO tend to invest in the same venture continuously instead of shooting once at numerous single ventures. The models of hypothesis 4 containing three regressions demonstrated the expected significant effects of LTO on the percentage of investments in the early and expansion stages [early stage: β = −0.068, t = −1.92, p = 0.056, and 95% CI = (−0.137, 0.002); expansion stage: β = 0.052, t = 2.22, p = 0.027, and 95% CI = (0.006, 0.098)]. However, LTO showed no significant effect on the percentage of investments in the mature stage [β = 0.015, t = 0.51, p = 0.608, and 95% CI = (−0.044, 0.075)]. This result suggested that VCs with higher level of LTO are more interested in investing in ventures in the expansion stage but more conservative in ventures in the early stage. In addition, they maintain a neutral attitude toward mature ventures. Thus, hypothesis 4 was partially supported.

In addition to the main findings, the author also observed some significant coefficients of the control variables. The industrial horizon showed negative effects on all the dependent variables except for the new ventures in the early stage, possibly because VCs pursuing industrial diversity of their investment portfolios may reduce their attention on other attributions of ventures, such as human capital and popularity. Moreover, in order to invest in a larger industrial range, VCs may need to separate their money into pieces to invest more ventures in different industries. Thus, they may lack concentration on single ventures and attempt to catch opportunities from the early stage with fewer expenses. Conversely, a larger network of VCs results in more investments in high-level human capital new ventures as well as new ventures in the extension and mature stages. This outcome may be due to the information and knowledge spillover from the other members of the network.

For robustness check, the author first ran the models with all variables to be standardized. The results turned out to be almost consistent. For H2, H3, and H4, which were supported in the main models, the author carried out extra models for robustness check.

For H2, which was about industrial popularity, the author applied a new measure to capture the industrial popularity level of the portfolios, which was similar to the measure of human capital explained in the Section “Materials and Methods.” The author calculated the mean of the popularities of industries yearly and determined a venture as popular if it belonged to an industry with higher popularity than the mean. Then, the author accessed the percentage of high-popularity new ventures by dividing the number of popular ventures the VC invested in by the total number of ventures it invested in the focal year.

For reinvestment (H3), the author calculated the time a venture invested before the focal year by the focal VC (if a venture was first invested by the focal VC in the focal year, it was equal to 0). Then, the author integrated them to the portfolio level by calculating the means across portfolio ventures of a VC on each focal year as the measure of the focal VC’s preference for reinvestment.

As for the hypothesis of preference on stages (H4), the author substituted the classified measures of stages with a continuous one. The author first coded the round by sequence: “Seeds” is equal to 0, “Angel” is equal to 1, “Pre-A/A/A+” is equals to 2, “Pre-B/B/B+” is equal to 3, “C/C+” is equal to 4, “D/D+” is equal to 5, “E” is equal to 6, “F to before-IPO” is equal to 7, and “after IPO” is equal to 8. Then, the author calculated the average round of each VCs’ portfolio ventures yearly, the higher of which indicated the tendency of a VC to invest in more mature ventures.

The author reports the additional results of the robustness check in Table 5. For H2 (industrial popularity) and H3 (re-investment), the significant coefficients of LTO again supported the two hypotheses [H2: β = −0.095, t = −2, p = 0.046, and 95% CI = (−0.189, −0.002); H3: β = 0.088, t = 2.95, p = 0.003, and 95% CI = (0.030, 0.147)]. One percent rise in the VC’s LTO may cause a 9.5 decline in the percentage of high-popularity ventures in its portfolio and a 0.088 increase in the frequency of investing in the same venture. For H4, which proposed that LTO is first positively related to preference in later stages and becomes negative after, the author tested the possible inverted “U”-shaped relationship between LTO and preference of the investment stage by introducing the squared term of LTO, as shown in Equation (2). Consequently, the inverted “U”-shaped relationship was supported, as the coefficient of LTO was significantly positive while the coefficient of the squared term of LTO was significantly negative [LTO: β = 1.125, t = 3.85, p = 0, and 95% CI = (0.551, 1.699); LTO^2: θ = −0.554, t = −2.42 p = 0.016, and 95% CI = (−1.003, −0.105)]. The inverted “U” shape is clearly visible in Figure 1. This outcome confirmed our hypothesis that VCs with higher LTO prefer ventures in the expansion stage over those in the early and mature stages.

This study is limited in several ways. First, the author did not test the moderate effects of other related factors. Evidence shows that affects and cognitions are always intertwined in the decision making process (Elen et al., 2013). Future research can contribute by testing whether their interplay still exists at the organizational level. In addition, environmental factors can serve as important contingencies (Nadkarni et al., 2016; Lin et al., 2019) despite the challenge that venture investors may react to the environment much differently from top executives, as they have to look across industries and quickly catch the dynamics. Another deficiency is that the author did not have the opportunity to test the effect of LTO on investment performance. Our approach takes advantage of emerging online media resulting in limited length of the observation period. The author was only able to collect LTO data in the past few years; hence, the time window to observe investment performance is insufficient, as it takes years to see whether the investment can be paid back (Drover et al., 2017). Nevertheless, whether LTO can bring VCs with better investment performance is a research topic worthy of investigation. Another rising force is CVC with dual motivations for industrial and financial goals. Investigation on CVCs will reveal resonant effects of the psychological mechanism on both the strategic and financial decisions compared with independent VCs focusing more on financial goals. Meanwhile, the approach of this study can also be applied to similar frameworks about start-ups to investigate the impacts of LTO or other psychological factors on performance at the organizational level. Lastly, studies on venture investment always treat the formation of investment ties as a unilateral decision, made by either investors or entrepreneurs. The proven similarity bias exiting in a venture investment suggests more extensive “matching” effects between ventures and investors. Hence, future researchers are encouraged to explore “matching” effects under psychological frameworks.