Using a Vector Autoregression (VAR) Model to Demonstrate How Forecasting Can Improve an Organisation's Understanding of its External Environment – An Empirical Investigation of the Health Sector

Introduction:The budgeting process is a primary function, to provide a level of control, for the management of companies by providing sales and costs benchmarks against which managers are required to perform. Budgets also play... [ view full abstract ]

Introduction:

The budgeting process is a primary function, to provide a level of control, for the management of companies by providing sales and costs benchmarks against which managers are required to perform. Budgets also play an important planning role, which practitioners now consider more important than the control function (Hansen, Otley et al. 2003, Hansen and Van der Stede 2004, Selto and Widener 2004, Brueggen, Grabner et al. 2014). Controlled based forecasts often examine predetermined benchmarks rather than understanding how the external environment is influencing the organisation (Bittlestone 2000, CIMA 2004, Cassar and Gibson 2008, Sivabala0n, Booth et al. 2009). The research objective of this study is to undertake an empirical investigation into the role of forecasting information in understanding the relationship between an organisation and its external environment. An alternative forecasting technique, Vector Autoregression (VAR) modelling, is used to establish shorter-term and medium-term forecasts in order to provide a deeper insight for management of how the performance of the corporation will react to a changing environment. Irish healthcare sector data is utilised. The healthcare sector in Ireland is a contextually rich setting for this analysis, particularly given the complexities and inherent market failures that exist in a health setting, making it difficult to make assumptions about the supply and demand for services. There are asymmetric information problems between the providers and consumers of services (Arrow 1950). Moral hazards also often exist where there is an incentive to over-consume free services and adverse selection applies as health insurance operates in the market (Zweifel and Manning 2000).  

This study seeks to understand the relationships between health expenditure, income and demographics. By understanding the dynamics between these variables, health policy makers will be able to forecast the impact that changes in economic and demographic factors will pose on the sector in a more informed way. For example, this research provides an understanding into the implications of increased health expenditure in response to a period of high economic growth. 

Research Methodology & Empirical Data

VAR models have become a benchmark-forecasting tool for understanding dynamic economic relationships. VAR models are a system of equations that regress each variable on a number of lags of that variable and the current and past values of all the other variables (Enders 1995).  They were first developed by Sims in 1980, who sought to address some of the limitations that were arising in single equations forecasting models (Sims 1980). Since VARs have been established, they have predominantly been used in the area of financial markets and monetary policy; however more recently their use is spreading into other areas such as health care policy (Jones, Evans et al. 2009, Lopreite and Mauro 2017).

The study uses health expenditures data in Ireland between 1970 and 2016 (OECD 2016, Department of Health 2017). Income has been established as a determinant in health expenditure (Newhouse, 1977). It has been also been highlighted that this market will be affected by changing demographics over the coming years and this is seen as a major risk for the delivery of services (Maev-Ann Wren 2017).  The forecasting model also includes three macro-economic variables, government health expenditure, GDP and an age index, which is the number of people over 65 years divided by the number of people of working age.  The VAR forecasting model estimated is presented in Figure 1.

This three variable VAR model allows impulse response functions to be created. Impulse response functions use a single graph to represent the response of each variable to an unexpected change in another variable. It is also possible to understand the inter-relationships between variables by decomposing the error variable to understand the proportion of the movement in a sequence due to its own shock compared to the shock of another variable.  

 Empirical Results – Preliminary Analysis & Discussion

The impulse response functions from this model are shown in Figure 2 and Table 1.  The top left column in Table 1 show the percentage change in health expenditure following a 1% unexpected move the other variables. Therefore, if GDP unexpectedly increases by 1% then health expenditure will increase by 4.6% after six years and 5.7% after ten years. Figure 2 highlights health expenditure responds within the first five years to an unexpected change in GDP. Over the same period, there is a much smaller response to an unexpected change in the age index. However, over a ten-year period, the response of health expenditure to a GDP shock slows, while the response to an age index shock is much larger over this period.

The decomposition of the error variance which is outlined in right columns of Table 1. This shows that over 60% of the error in government health expenditure can be explained by external factors six years after the shock. Ten years after a shock 75% of the variance in health expenditure is explained by external factors.

These findings highlight the need for policy makers to view health expenditure forecasting over a longer period.  For example, by responding on an annual basis to the government finances, the Department of Health is failing to understand the medium term impact that economic cycles have on health expenditure. As the population ages over the next few decades, there are longer-term healthcare needs that will need to be financed, regardless of the stage of the economic cycle. Failure to appropriately plan for this could result in a shortfall in the provision of health services in Ireland, notably if the effects of a recession were to take place when the effects of the ageing population were also impacting. 

Conclusion and Policy Implications:

This study provides a deeper insight into how an organisation responds to its external environment. This technique helps to inform management how an unexpected change in a factor outside of their control, may impact and hence they can respond in a more informed way. 

The study also contributes to the management accounting in healthcare literature by focusing on how forecasting information can led to improved resource allocation decisions. It highlights the need for longer term planning, in order to avoid a potential disruption to health services if an recession were to occur when the effects of the ageing population were impacting.

Government health expenditure, is highly dependent on its external environment. For example, the government financing of health services is determined by economic cyclical factors, even though the demand for services will be consistent through economic cycles

References:

Arrow, K. J. (1950). "A difficulty in the concept of social welfare." Journal of political economy 58(4): 328-346.

Bittlestone, R. (2000). "Tackling the numbers game." Accountancy 125(1278): 46-46.

Brueggen, A., I. Grabner and K. L. Sedatole (2014). "The Folly of Forecasting: The Effects of Sales Forecast Accuracy and Bias on Inventory and Production Decisions under Aggregated and Disaggregated Forecasting Regimes." Journal of Management Accounting Studies.

Cassar, G. and B. Gibson (2008). "Budgets, internal reports, and manager forecast accuracy." Contemporary Accounting Research 25(3): 707-738.

CIMA (2004). A report in better budgeting a report on the better budgeting forums from CIMA and ICAEW, Institute of Chartered Accountants in England & Walse. .

Department of Health (2017). Health in Ireland, Key Trends 2017. Annual Dublin, The Goverment of Ireland

Enders, W. (1995). Applied Econometric Time Series. New York, Wiley.

Hansen, S. C., D. T. Otley and W. A. Van der Stede (2003). "Practice developments in budgeting: an overview and research perspective." Journal of management accounting research 15(1): 95-116.

Hansen, S. C. and W. A. Van der Stede (2004). "Multiple facets of budgeting: an exploratory analysis." Management accounting research 15(4): 415-439.

Jones, S. S., R. S. Evans, T. L. Allen, A. Thomas, P. J. Haug, S. J. Welch and G. L. Snow (2009). "A multivariate time series approach to modeling and forecasting demand in the emergency department." Journal of biomedical informatics 42(1): 123-139.

Lopreite, M. and M. Mauro (2017). "The effects of population ageing on health care expenditure: A Bayesian VAR analysis using data from Italy." Health policy 121(6): 663-674.

Maev-Ann Wren, C. K., Brendan Walsh , Adele Bergin , James Eighan , Aoife Brick ,  Joanne Banks, Sheelah Connolly , Dorothy Watson (2017). Projections of demand for healthcare in Ireland, 2015-2030: First report from the Hippocrates Model. E. a. S. R. Institute. Dublin, Ireland, ESRI.

Newhouse, J. P. (1977). "Medical-care expenditure: a cross-national survey." The Journal of Human Resources 12(1): 115-125.

OECD (2016). "Health at a Glance 2015 - OECD Indicators - en."

Selto, F. H. and S. K. Widener (2004). New directions in management accounting research: insights from practice. Advances in Management Accounting, Emerald Group Publishing Limited: 1-35.

Sims, C. A. (1980). "Macroeconomics and reality." Econometrica: Journal of the Econometric Society: 1-48.

Sivabalan, P., P. Booth, T. Malmi and D. A. Brown (2009). "An exploratory study of operational reasons to budget." Accounting & Finance 49(4): 849-871.

Stock, J. H. and M. W. Watson (2001). "Vector autoregressions." The Journal of Economic Perspectives 15(4): 101-115.

Zweifel, P. and W. G. Manning (2000). "Moral hazard and consumer incentives in health care." Handbook of health economics 1: 409-459.