Impact of Global Purchasing and Supplier Integration on Product Innovation

The data used in this study are from the International Purchasing Survey (IPS), an international online survey on purchasing and supply management conducted in 2009. The survey covers complete answers from 679 manufacturing firms (ISIC codes 25-30) in Europe, USA and Canada, and the data were collected by a network of partner universities. Several papers have previously been published that are based on the IPS. The questionnaire included questions on firms' purchasing strategy, practices and performance, and was mainly answered by senior purchasing managers or the equivalent. Six-point scales were used in instances where it was important to avoid a neutral middle response. In instances where avoiding a neutral middle response is not critical, e.g. when rating performance, seven-point scales were used. The responding company answered questions at both the firm level and the component category level. The respondents were asked to choose a category that was relatively homogenous in terms of products and services. The data were analyzed by comparing means (t-test) as well as factor and regression analysis, using SPSS software.

Global purchasing

This study focuses on the relative characteristics of global purchasing and distinguishes between two groups: global and regional. The geographical area of the firms' purchasing activities in combination with the respondent's home country produced the two categories, where regional means purchasing is only done within the same continent, i.e. Europe or North America. An American company purchasing from the USA and Canada and Mexico would thus be considered regional, as would a German company purchasing from, say, UK, Hungary and Italy. Any company buying from China or Australia would be considered to be engaged in global purchasing. The scope of global purchasing was measured by asking from which of the 13 predefined geographical areas the firm buys more than 10 per cent of a selected category of goods and services (Table I). The threshold of 10 per cent was selected to disregard one-of-a-kind contracts and focus on major purchasing patterns. A relatively low percentage is appropriate, since some innovative firms may purchase the most innovative components globally, while purchasing the majority locally in order to minimize supply risks. It is also possible that low-cost supplies of relatively less innovative components are only available in faraway countries, whereas the most innovative components in a category are purchased locally. The 10 per cent threshold shows that a company is indeed considering suppliers on a global scale, while the majority of actual purchasing does not need to be global, in line with Quintens et al. and Mol et al. In other words, this study measures the scope of global purchasing, but not the depth. Based on this we could define two groups, regional and global, of firms among the 679 respondents (Table I).

Priorities, control variables and performance

The question to measure priorities for purchasing was phrased "Please indicate to what extent management has emphasized the following priorities for the chosen category over the past 2 years". The priorities were measured on a six-point Likert scale, from "not at all" to "completely". Two priorities were related to innovation and thus relevant for this study (Table IV). The priorities are used for testing H1a and H1b, but are not included in further analysis, as it is, in this context, uninteresting to analyze whether priorities correlate with performance improvement within the same area.

A typical control variable in the operations management literature is company size, usually measured as turnover. As this paper is concerned with the purchasing department, the first control variable is accordingly the turnover of the department or total purchasing spending (Table II). Since this is an international survey, some currencies needed to be converted to euros. The exchange rate used was the average exchange rate during 2010 obtained from the Central Bank of Sweden.

The most obvious contribution to innovation from suppliers occurs when the company selects suppliers that are contributing something unique and innovative. The second control variable is according to what extent suppliers provide access to unique assets or resources, and was measured on a six-point scale from "extremely low" to "extremely high". By introducing the control variable, the direct effect of having innovative suppliers is removed, and the analysis can focus on how the purchasing department can leverage the innovative potential of suppliers.

As previously mentioned, two performance indicators were used: the supplier TTM for new or improved products or services and the level of innovation in products or services from suppliers. As was previously explained, these two variables may be conflicting. It is thus appropriate to use them individually, instead of combining them in one factor. The question to the respondent was phrased: "Please consider current category performance – compared to management targets – for the following objectives". Seven-point Likert scales were used, ranging from much worse than target (1) to much better than target (7). The descriptive statistics for these are displayed in Table II.

Table II. Descriptive statistics for control variables, priorities, market characteristics and performance

Supplier integration

Supplier integration was measured by letting the respondent indicate with what frequency a number of integration tools are used to support the purchasing activity and the relationship with supplier(s) for the chosen category, on a six-point scale ranging from "never" to "always". The tools were identified by a team of international scholars as established tools used by practitioners, and are displayed in Table III. These tools, also in line with Flynn et al., are primarily aimed at improving the flow of goods and not specifically NPD.

Not measuring supplier integration into NPD can be seen as a limitation of this study. Overcoming this limitation would probably require involving departments other than purchasing in the data collection. However, previous studies have shown that the use of tools aimed at operational integration has an impact on performance, including innovation. One reason for this may be because such tools afford extensive information sharing with suppliers, information that can also be utilized in the NPD process. For example, sharing production planning or inventory levels will ensure availability of components for a new product, thus supporting TTM. When the product development process speeds up, companies are also likely to be able to manage more processes, leading to a higher aggregate level of innovation, although some argue that the two cannot be improved simultaneously and must be traded off.

An exploratory factor analysis was conducted for the six items for two reasons. First, we wanted to reduce their numbers to assist in further analysis. Second, the high degree of convergence between the items, as shown in Table III, would cause multicollinearity problems if treated separately. All items loaded onto a single factor with high factor loadings and a high Cronbach's α value, implying a high degree of construct validity. We refer to this factor as "supplier integration" from here on.

Table III. Exploratory factor analysis of supplier integration

Proficiency in supplier integration

With the term proficiency in supplier integration, we refer to purchasing skills in key purchasing processes aimed at integrating suppliers. The purchasing proficiency concept builds partly on the framework of González-Benito, which in turns builds on Vickery's theory of production competence. This study does not cover early purchasing activities such as finding and selecting suppliers, since those activities were presumably conducted when the firms decided to purchase regionally or globally. In contrast, we are concerned with how the established suppliers are managed. The main focus is on a logistics type of integration, since NPD process is typically beyond the scope of the purchasing department. The construct contains one item on supplier involvement in the NPD process to take into account pre-NPD activities such as providing specifications, evaluating suppliers or signing contracts related to NPD projects. The other two items (Table IV) concern whether the suppliers are involved in the ordering process. Purchase orders are usually generated electronically, using an MRP or ERP system, which requires integrating suppliers. Integrating suppliers electronically requires a high degree of proficiency, since detailed, error-free descriptions, giving the technical details of the purchased product, unit price, delivery time, etc., are required.

Table IV. Exploratory factor analysis of proficiency in supplier integration

Proficiency was measured by letting the respondent indicate the level of proficiency (i.e. the level of quality in executing each process) for the chosen category, on a six-point scale ranging from "extremely low" to "extremely high". Exploratory factor analysis was also conducted for these items, for the same reasons as the supplier integration construct. All items load onto a single factor, which assures high construct validity. The factor is used in subsequent analysis and referred to as "proficiency in supplier integration".

Summary of variables

The analysis will thus consist of two independent variables: "Supplier integration" and "Proficiency in supplier integration", and two control variables: "Purchasing volume" and "Extent to which a supplier provides access to unique assets and resources". Two dependent performance variables are used: "Supplier TTM" and the "Level of innovation sources from suppliers". Table V displays the correlation (Pearson) between all independent and dependent variables, and shows, as expected, correlations between most independent variables and dependent variables. It also shows correlations between some independent variables, which can indicate potential multicollinearity problems. However, subsequent analysis (see end of next section) shows that the variance inflation factor (VIF) is well within safe levels.

Table V. Correlations