Novartis

Measuring and valuing the social impact of Wages – The Living Wages Dataset and The Health Utility of Income

About the Company

Novartis International AG is a Swiss multinational healthcare company that provides solutions to address the evolving needs of patients worldwide.  At the end of FY17, the organization recorded about 126,000 employees and $49 billion in net sales. Novartis applies science-based innovation to address some of society’s most challenging healthcare issues.

About the Approach

The concept of living wage developed long ago by the ILO and other organizations, is a key method to assess wages. The living wage should provide a satisfactory standard of living to the workers and their families. Recent developments in this field have established a global framework to measure it. Receiving a wage below a living wage threshold will ultimately lead to a reduced quality of life and life expectancy. On the contrary, receiving a wage above the living wage threshold will positively influence your life quality and expectancy. Some datasets have shown that differences in life expectancy can reach 13 years in the US for instance between the low- and high-income population group, and 13 years in France.

Given the rising interest of the private sector (and other sectors) to measure its value and impact beyond traditional financial metrics, Novartis in collaboration with Valuing Nature Network have developed a global dataset of living wages per country as well as a measure of the human health impact of wages, called the Health Utility of Income (HUI) to measure and value the social impact of wages worldwide. The Living Wage Dataset is believed to be the first to be publicly published covering almost all countries in the world.

Stage 1: Frame

Get Started

Social impacts, both negative and positive, are key elements of the Financial, Environmental and social (FES) impact valuation, the Novartis version of the Triple Bottom Line Approach.  

Stage 2: Scope

Define the objectives

In the quest to measure and value the social impact of wages, the global dataset of living wages is intended to support Novartis and other companies to benchmark the wage levels provided in their direct operations as well as through their value chains.

The Health Utility of Income (HUI) will support businesses in valuing their impact on society, often referred to as impact valuation, and in line with the Natural and Social Capital Protocols.

Scope the assessment

Global Dataset of living Wages: The objective was to calculate such a dataset for all countries, despite the data limitations and despite the fact that an average living wage per country will only be a first approximation and will need to be refined for specific locations within a country.

Health utility of Income:  In order to develop the HUI model, the objective was to collate data from all countries in the world.

Determine the impacts and/or dependencies

The global dataset of living wages and the Health Utility of Income are part of the same impact pathway; however, they play different roles. The figure below illustrates the social impact of employment through wages (or income). The latter is considered as an output of employment, yet it could generate a positive or negative impact. Only a comparison with a living wage could provide this insight. The difference between the provided wages and living wages indicate the level of income inequalities, which in turn generates health inequities. In order to translate income inequalities into health inequities, the Health Utility of Income factors are required. An illustration of the use of the two data sets is provided below.

Stage 3: Measure and Value

Measure impacts and/or dependencies 

A. Global Dataset of Living Wages

The Living Wage Global Dataset (LWGD) features estimates based on locally-relevant data rather than macro-economic models. The hierarchy is as follows:

  1. Living wage estimates at country level when it exists already (bottom-up)
  2. Data that allows to calculate living wages (bottom-up)
  3. Macro-economic model (top-down)

The approach used two main data sources for the bottom-up part of the LWGD. On the Wage Indicator website, the Global Labor Organization (GLO) published an estimate of living wages for 57 countries based on proprietary surveys. No other significant dataset was found despite stand-alone initiatives, such as an estimate for the UK (D’Arcy and Finch 2016).

To cover more countries, an additional dataset was used from Numbeo, a website which collects the prices of different items in all countries around the world. The data collected is still limited and does not allow calculating a living wage and as such, other data sources and assumptions were used. This bottom-up model based on Numbeo data allowed to cover an additional 47 countries, on top of the 57 already covered in the dataset from the GLO.

In order to still provide an estimate for the countries not covered, a top-down approach was developed, which consists of extrapolating the living wage estimates using a model that correlates existing data points to one or more parameter(s). In this case, Novartis and Valuing Nature used the Purchasing Power Parity (PPP) from the Word Bank (2016) or more precisely a derived index based on PPP called “Price level ratio of PPP conversion factor (GDP) to market exchange rate.” The PPP parameter considers the fact that USD1 spent in the United States will not buy as many goods as the same US1 spent in Mali or China. Using PPP allows to compare local price levels for different countries.

The correlation indicated a coefficient of determination (R2) of 0.749 which is considered a good correlation, although further improvements could be considered. This top-down model allowed to cover an additional 76 countries on top of the 104 already covered by the bottom-up approach, reaching 180 countries in total. The figure below provides an overview of the modeling approach for creating the living wage global dataset. It highlights the prioritization level which indicates the different sources of data from bottom-up (step A and B) to top-down (step C). An overview of the living wage global dataset modeling is included below.

The dataset ensured consistency of scope and modeling across different data sources, in particular regarding typical family composition and working family members. The dataset provides as well low and high estimates for each country covered, to address the difference that exists between living wages in urban and rural areas.

B. Health Utility Income (HUI)

The impact valuation of employment or income is not well developed in the field of social capital accounting in the corporate sector. Traditional economic approaches have been used in many social capital accounting applications by a number of large corporations. These approaches usually consider that all wages are generating a positive value. The valuation is based on the real financial value and usually applied in combination with a multiplier to account for the benefit of the first job created and for the indirect jobs created or sustained by this first job. A threshold and other multipliers have been used to account for the local context and different levels of wages, but still based on the assumption that every job created is a good one.

In this work, an innovative model was developed based on the social determinants of health and the concept of marginal utility of income. Social determinant of health studies have been widely developed among others by the World Health Organization. The basic idea is that our health (well-being) and life expectancy are defined by a variety of determinants, including behaviors, environmental factors, genetic disposition and of course social determinants. Among the social determinants, we can find early age development, education, work environment and conditions, retirement conditions and social protection/services. Working conditions, including incomes, play an important role but are only part of the picture.

In order to develop the HUI model, the data was collected from Eurostat and OECD data which link life expectancy to education level, a good proxy of income. Isolated data exists for some countries on the direct link between income and life expectancy, for example for France.

Public data does not exist for all countries in the world. To overcome the data gap, a model was built to estimate the health inequities related to income inequalities worldwide. When measured along income inequalities, health inequities are called a “social gradient”. A social gradient links health inequities (for instance life expectancy) with income differences. In short, people who are less advantaged in terms of socioeconomic position have worse health (and shorter lives) than those who are more advantaged.

The results provided the social gradients for all countries, expressed in DALYs per year of work. This DALY can also be expressed in monetary units using different valuation techniques. We used the productive value of life, equivalent to approximatively USD 44’000/DALY, which is based on the OECD average GDP per capita. The social gradients calculated still need an extra step of modeling to be translated into Health Utility of Incomes.

Measure changes in the state of social & human

From health inequities to the “health utility of income” (HUI)

We are departing from the assumption that the living wage, as described earlier, is the threshold below which a negative impact can be observed and above which a positive impact occurs. This threshold has been recommended by other companies within the field of social capital accounting such as Nestlé and DSM. Setting a baseline is relatively subjective and is a choice that needs to be agreed upon. Depending on the income a person receives, and its relative difference to the baseline, the health inequalities will vary. For a person earning USD 100’000 /year, the benefit of earning an additional USD 5’000 will be less than for a person earning USD 15’000 /year. This is why we cannot just apply the health inequities data at a flat rate for all wages paid by a company. This is the concept of marginal utility of income. This relationship has been documented for other countries like the USA, Australia, etc. Trends across countries are relatively similar. The marginal utility falls to a low level above an income representing 3 or 4 times the living wages, as illustrated in below.

We allocated the health inequity data into different income gaps, below and above the living wage baseline, as illustrated in Table 1 following a typical curve of utility of income. According to this model, the health inequities linked to income inequalities are almost inexistent above a level of income which is four times the living wage. The resulting factors obtained are called the Health Utility of Income (HUI) and are the main factors used in the impact valuation model for Novartis.

Value impacts and/or dependencies

Novartis used the living wage dataset to benchmark salaries for the societal impact of living wage valuation. The analysis included all salaries provided to Novartis employees around the world, which are more than 120’000. Novartis has already implemented a living wage policy to ensure no employee is paid below a living wage and adjust any salary where deviations from the policy occur. The new Living Wage Global Dataset provides a new point of comparison that Novartis can use for own operations and its supply chain. Based on the global living wage dataset and HUI approach, the social impact of living wages was calculated for 75 countries in which Novartis operates, in 2017.

Primary data on wages and salaries in the supply chain is not readily available. However, this is required to extend the social impact of living wages to the supply chain and inform about the quality of jobs created through the purchase of goods and services. To extend this assessment to the supply chain of Novartis, we used spend data linked to an input-output model providing socio-economic indicators (Exiobase) to estimate the number of indirect jobs linked to Novartis. The input-output model also allowed classifying the number of jobs per skill level, which was used to define the salary levels to be used with the HUI factors. This was done for a total of 167 countries.

According to our economic impact analysis for 2017 based on the World Input-Output database, Novartis spend generated 360´000 jobs in its supply chain. The Exiobase analysis suggests the following skill distribution: 65% high-skilled jobs, 14% medium-skilled jobs and 21% low-skilled jobs. Jobs were identified per sector, but more importantly for our model, per country as this informs the reference levels of wages. The wage levels were defined using the World Income Inequalities Database (WIID). For low-skilled jobs, we used the average income considering quintiles 1 and 2. For medium-skilled jobs, we used the quintile 3, while for high-skilled jobs we considered quintile 5.

Overall, the impact of Novartis supply chain generated a USD 5.6bn net positive societal impact, with a negative impact of USD 0.4bn and a positive impact of USD 6.0bn. The low-skilled jobs were estimated to have a negative social impact value of USD 0.2bn and a positive social impact value of USD 0.2bn. The medium-skilled jobs created contributed negatively with USD 0.2bn and positively with USD 0.3bn. The high skilled jobs contributed only positively with USD 5.5bn.

The direct employment impact of Novartis generated a positive impact value of USD 1.0bn, while seeing a very small negative impact of USD 1.4m (less than 0.2%) linked to the cases of salaries below the living wage threshold. These cases are considered temporary and were followed-up in accordance to the Novartis living wage policy.

The top ten countries with the highest social impact in the Novartis supply chain were not where the highest spend was found. Those countries are (in bracket the spend ranking): USA (1), China (6), Singapore (16), Germany (3), Switzerland (2), Brazil (18), Indonesia (43), United Kingdom (4), Austria (5) and Spain (11). There were 21 countries for which the social impact of living wages was positive for both the Novartis supply chain and Novartis operations. All of them were in Europe.

Negative impact values for low-skilled jobs in the supply chain were found in India (77% of the total negative impact driven by the IT and the chemical sector), China (11%), Brazil (3.2%), South Africa (2.5%), Turkey, Iceland, Singapore, Ukraine, Indonesia and Mexico (top 10).

Stage 4: Apply and Integrate

Take action 

The social impact of living wages has provided important insights into social hot sports in the supply chain and will inform strategic interactions with the suppliers.

The analysis was conducted for the first time on 2017 data. The results were used to raise the awareness internally on the social impact of living wages and on the identified hotspots.

The approach and results were shared externally, leading to exclusively positive and forward looking responses.

Reducing the hurdles for other companies to develop similar approaches is key for mainstreaming social and human capital measurement and management. Novartis uses various platforms to share its Financial, Environmental and Social impact valuation approach.