Econometric Approach to Financial Analysis, Planning, and Forecasting

Abstract

Following Chap. 4, we apply simultaneous equation models in discussing how econometrics methods and accounting data can be used for financial analysis, planning, and forecasting. The models used in this chapter include single-equation model, two-stage least squares model, three-stage least squares model, and SUR estimation method. In addition, we discuss the relationship among programming, simultaneous equation , and the econometric method.

This chapter an update and extension of Chap. 26 of the book entitled, Financial Analysis, Planning and Forecasting by Lee et al. (2017).

Appendix: Johnson & Johnson as a Case Study

1.1 Introduction

The purpose of this appendix is to use Johnson & Johnson’s annual data as an example to show how Spies’ model can be use to analyze an individual firm’s dynamic capital budget decisions. Firstly, Johnson & Johnson’s (J&J) operations are briefly reviewed. Secondly, both the balance sheet and the income statement for J&J during the period of 1997–2006 are used to evaluate its financial performance. Thirdly, both the endogenous and the exogenous variables needed to estimate the equation system. Implications of these regression results are also briefly analyzed.

1.2 Study of the Company’s Operations

Johnson & Johnson was incorporated in New Jersey on November 10, 1887. The company is engaged in the manufacture and sale of a broad range of products in the healthcare and other fields in many countries of the world. J&J’s worldwide operations are divided into three industry segments: consumer, pharmaceutical, and medical devices and diagnostics.

Consumer

Consumer products encompass baby and childcare items, skincare products, oralcare products, woundcare products, and women’s healthcare products.

Pharmaceuticals

The pharmaceutical sector includes products in the following areas: antifungal, anti-infective, cardiovascular, contraceptive, dermatology, gastrointestinal, hematology, immunology, neurology, oncology, virology, pain management, psychotropic, and urology fields.

Medical Devices and Diagnostics

The medical devices and diagnostics segment includes suture and mechanical wound closure products, surgical equipment and devices, wound management and infection prevention products, interventional and diagnostic cardiology products, diagnostic equipment and supplies, joint replacements, and disposable contact lenses.

Table 5.9 shows that all three divisions of the company are quite profitable and its product lines well-diversified. One should also note that the international division contributes as much as the domestic division, making the company less susceptible to the ups and downs of the US economy.

Table 5.9 Sales in different segment

Research activities are important to J&J’s business and account for about 14% of sales. The company employs about 122,000 persons worldwide engaged in the research and development, manufacture, and sale of a broad range of products in the healthcare field.

1.3 Analysis of the Company’s Financial Performance

Tables 5.10, 5.11A and B show J&J’s balance sheet and common-size income statements for the period 1997–2006. Tables 5.12 and 5.13 analyze the company’s performance as reflected in key ratios, trends, and the DuPont analysis.

Table 5.10 Balance sheet

Table 5.11 Common-size income statement ($ in millions)

Table 5.12 Ratio analysis

Table 5.13 DuPont analysis

As can be seen from Table 5.12, the company has a fairly comfortable current ratio and quick ratio. However, since 2006 the ratios have been declining slightly. On examining the receivables and inventory turnover ratios, one notes that the receivables turnover ratio has been increasing over the same period, implying that J&J is shortening the periods of credit to generate sales. Inventory turnover has fluctuated between 8.3 times and 12.7 times over the ten-year period. In other words, the average raw materials and finished goods inventory ranged between one to one-half months’ sales, which show that the firm does not carry a high inventory of slow-moving stocks; it also holds sufficient inventory to avoid stock-outs. On the overall, J&J seems to have a fairly comfortable working-capital and short-term liquidity position.

The long-term leverage position can be analyzed by examining the debt–equity ratio. From Table 5.12, one can see that, while this ratio has been rising since 1999, it is trying to lower its financial leverage which is around 5% in 2006. While this shows very low financial risk, it could also mean that the company is not taking advantage of financial leverage, especially since the company does not have much business risk, as will be seen in the DuPont analysis.

An analysis of the various trends show that sales have grown over twice and the growth has been very steady over the period. In fact, sales have been growing ever since 1950. Profits have kept pace with the sales growth, which is a healthy sign. Sales have grown faster than gross investment and working capital, implying that J&J is utilizing its capacity and working-capital funds better. A cursory glance of the balance sheet shows that the lower working-capital growth is due to the increase in current liabilities, which grew four times since 1997. Since the company has the funds to repay its account payables if necessary, this shows that J&J’s creditors are granting it longer periods of credit, showing increasing faith in the company’s stability.

Net worth (shareholders’ equity), sales, and profits have grown similarly, and this is because the company has been paying stable rate of dividends as reflected in the payout ratio in Table 5.11 and the dividend trend in Table 5.12. The formulas at the end of Table 5.13 show that the DuPont analysis is divided into three sections:

• 1. Operating performance as reflected in the asset turnover, gross profit margin, and operating leverage,

• 2. Financial efficiency as reflected in the financial leverage and assets/equity ratios,

• 3. Growth as shown in the retention ratio.

Asset turnover has been decreasing during 1997–2006. Gross profit margin has been fairly steady. However, the operating leverage as reflected in the EBIT/gross profit ratio has been increasing since 1973. This has been offset by the decrease in asset turnover, resulting in a steady return on investment (EBIT/total assets).

The company’s net profit/EBIT ratio has been improving, and this is because of decreasing debt leverage as described earlier. The assets/equity ratio has fluctuated slightly over the period. The combined impact of both these ratios resulted in J&J’s return on equity (net profit/shareholders’ equity) does not change too much in the last ten years. The company’s retention ratio has also been fairly stable, as was also seen earlier. The overall effect is that the ratio of retained earnings to equity has been fairly steady over the last ten years, with a slight fall in 1998 and 2001.

The combined impact of all three aspects shows that the company has a very steady operating performance with improving financial efficiency.

The final aspect of the analysis of the company’s financial performance is a comparison with the industry and the stock market as a whole. Value Line’s comparison of J&J’s rankings with the market as a whole is given in the following table:

Category	Rank
Timeliness Safety Financial strength Stock-price stability Earnings predictability Beta	3 1 A++ 95 100 0.60

1. Note Based upon value line report on November 30, 2007

From the above study of J&J’s operations and performance, one can see that the company has an excellent track record, especially as evidenced by the industry and overall rankings shown above. Further, the hospital supplies industry to which J&J belongs is a very steady industry and not affected by cyclical influences and other factors like changes in consumer’s tastes and preferences. While research and development is an important area in this industry, the technological developments are not so rapid to make this industry as volatile as some others, like the electronics industry.

This historical study has been restricted mainly to the last ten years because the observations made in this section will be used to supplement the analysis made using Spies’ model. One should also note that many of the comments made on J&J’s performance in the 2000s were also applicable to earlier time periods.

1.4 Variables and Time Horizon

The variables used in the empirical testing of the model were exactly the same as those used in Spies’ model. The following section explains the methodology followed in computing the various variables used in the model.

• Dividends (DIV). Only equity dividend was considered in the computation of this variable. Preferred dividend was paid only up to 1955, and therefore, for consistency’s sake, preferred dividend was deducted from cash flow for the period and not added to equity dividends. Also, stock dividends were not considered in the computation.

• Short-Term Investment (IST). Short-term investment is the net change in the corporation’s holdings of current and other short-term assets.

• Long-Term Investments (ILT). Long-term investments is defined as the change in gross long-term fixed assets and noncurrent marketable securities.

• Debt Financing (DF). This component is simply the net change in the corporation’s liabilities, both long-term and current.

• Equity Financing (EQF). The change in stockholders’ equity, minus the amount due to retained earnings, was used as the definition of equity financing. Though no new shares were issued to the public over the period studied, adjustments were made for stock splits, changes in common stock in treasury, and stock issued to employees under options exercised and stock compensation agreements.

• Cash Flow (Y). This variable was calculated by adding depreciation to net profit for the period and adjusting for other noncash entries in the retained-earnings statement as well as for preferred dividends paid.

• Corporate Bond Rate (RCB). This variable was not included in the model, since the corporate bond rate is not available in Compustat.

• Debt–Equity Leverage (DEL). The debt–equity ratio at the beginning of each period was computed for this variable.

• Dividend–Price Ratio (RDP). For this variable, the average dividend–price ratio for the period was used.

• Rate-of-Return (R). The ratio of the change in earnings to long-term investment in the previous period was taken as a measure of the rate-of-return on investments.

• Capacity Utilized (CU). Since J&J is a multiproduct company, the ratio of sales to gross fixed assets was taken as a proxy for the capacity utilized.

• Time Horizon. Annual data were used in the empirical study and the time period covered was 1950–1979.

Tables 5.14A and B list the data collected for the variables above. The data for 1969 have been omitted from the study because J&J consolidated its accounts that year, to include its foreign subsidiaries and this distorted the sources and uses of funds for that year.

Table 5.14 Endogenous and Exogenous Variables of Johnson & Johnson during 1949–1978

1.5 Model and Empirical Results,

The model used in the empirical study was the regression model developed by Spies (1974). Following Eq. (5.16) of the text, the model is defined as

$$ X_{t} = BX_{t - 1} + CZ + U_{t} , $$

(5.16)

where

• $X_{t}$ is a 5 × 1 matrix with the following variables:

• $$ X^{\prime}_{t} = ( {\text{DIV}}_{t} \quad {\text{IST}}_{t} \quad {\text{ILT}}_{t} \quad - {\text{DF}}_{t} \quad - {\text{EQF}}_{t} )^{\prime } ; $$

• $Z_{t}$ is a 6 × 1 matrix, which includes a constant (I) where

• $$ Z^{\prime}_{t} = (1\quad Y_{t} \quad {\text{RDP}}_{t} \quad {\text{DEL}}_{t} \quad R_{t} \quad {\text{CU}}_{t} )^{\prime } ; $$

• B is a 5 × 5 matrix of coefficients and C is a 5 × 6 matrix of coefficients;

• $U_{t}$ is a 5 × 1 matrix which represents error terms.

Since annual data have been used, no dummy variables are used to remove seasonality.

The “sources-equals-uses” identity implies that $\Sigma _{i} X_{i,t} = Y_{t}$ must be true for every period t; here, $X_{i,t}$ is defined as the ith variable in the X-matrix at period t. In order to incorporate this into the estimation procedure, the estimators must be restricted in such a way that, across equations, the coefficients of Y add up to 1, while the coefficients of the other exogenous variables add up to zero. Algebraically, these constraints can be stated as:

$$ \sum\limits_{i = 1}^{5} {\hat{c}_{i2} } = 1\quad {\text{and}}\quad \sum\limits_{i = 1}^{5} {\hat{b}_{ij} } = \sum\limits_{i = 1}^{5} {\hat{c}_{ik} } = 0\,{\text{for all }}j{\text{ and all }}k \ne 2, $$

where $b_{ij}$ is a coefficient in the B-matrix corresponding to the ith row and jth column and $c_{ik}$ corresponds to the ith row and kth column in the C-matrix, and $c_{i2}$ refers to the column for the variable $Y_{t}$ as discussed in Sect. 5.3 of the text.

Equation (5.16) defines five multiple regressions, which are used to describe the behavior of five endogenous variables , DIV, IST, ILT, DF, and EQF. Exogenous variables for these multiple regressions are Y, DEL, RDP, R, and CU.

Based upon data listed in Tables 5.14A and B, both OLS and constrained SUR methods are used to estimate the coefficients of five multiple regressions. OLS regression results are listed in Table 5.15A, and the SUR results are listed in Table 5.15B. There are 10 OLS regression coefficient estimates that are significantly different from zero under 10% significance. However, there are 14 constrained SUR coefficient estimates that are significantly different from zero under the same significance level. This implies that the efficiency of the constrained SUR method is higher than that of the OLS method. In addition, the results in Table 5.15B indicate that the empirical results for ${\text{DIV}}_{t}$, ${\text{DF}}_{t}$, and ${\text{EQF}}_{t}$ are more significant than those for ${\text{IST}}_{t}$ and ${\text{ILT}}_{t}$. If quarterly instead of annual data are used to estimate these five equations, we would expect that the performance of empirical results would improve substantially. Considering that OLS and SUR methods neglect information contained in the other equations, we use two-stage least squares method to estimate the coefficients of five multiple regressions. Similar to SUR method, Table 5.15C shows that the 2SLS results for DIV_t, DF_t, and EQF_t are more significant than the 2SLS results for IST_t and ILT_t.

Table 5.15 A OLS estimates for the period 1950–2006. B Constrained SUR estimates for the period 1950–2006. C Two-stage least square estimates for the period 1951–2006

Finally, Spies (1974) indicated that Eq. (5.16) can be used to predict $X_{t}$ given ${\text{Z}}_{t}$, and ${\text{Z}}_{t - 1}$. The model used to do one-period forecasting can be easily derived from Eq. (5.16) as

$$ \begin{aligned} X_{t - 1} & = BX_{t} + CZ_{t + 1} + U_{t} \\ & = B^{2} X_{t - 1} + BCZ_{t} + CZ_{t + 1} + BU_{t} + U_{t - 1} . \\ \end{aligned} $$

(5.17)

The applications of Eq. (5.17) to forecast are left to students themselves by using the data in Tables 5.14A and B.