1 Accounting Value Relevance in Turkey: Controlling for Shifting ...

1

Accounting Value Relevance in Turkey: Controlling for Shifting Accounting Standards, Inflation and the GFC

Abstract

This paper examines accounting-stock price value relevance in the developing Turkish market using an extended Ohlson model. We find (i) incremental explanatory power from the exogenous impact from inflation and real interest rates, (ii) a continued increase in value relevance for book values after the introduction of inflation accounting and IFRS, and (iii) a strong reversal in the earnings coefficient from positive directly before the GFC, to negative during and after the GFC. Questions are asked as to whether the pre-GFC market based accounting adjustments were effective in communicating fundamental valuation information, and the role accounting plays in transition economies.

Keywords: Turkish inflation accounting, Accounting valuation and the GFC, Accounting valuation in emerging economies

2

1. INTRODUCTION

There is an abundance of papers that assess changing value relevance after the introduction of

International Financial Reporting Standards (IFRS) in developed markets, especially Western

Europe (see Bruggemann et al., 2013, for a review). However, research on accounting

valuation impacts in transition markets is sparse, with little known outside the stock markets

of the U.S./Western Europe/Australia.

The introduction of contemporaneous and conceptually based accounting systems

(such as IFRS) in transition/developing markets promises higher quality information feedback

to market actors. This is because these markets suffer from asymmetries resulting from a

shortage of information facilitators (such as trained financial analysts), and lower corporate

governance standards and higher corruption levels that inhibit outside information flow. In

this regard, Filip and Raffournier (2010) argue, in emerging economies, accounting is the

primary source of information for investors with higher quality imposing a form of external

governance. Overlying these information based problems, is the mitigating influence from

uncommonly high levels of inflation that potentially compromise the valuation role of

financial accounts (Edwards and Bell, 1961). All these factors converge to raise the

importance of a study into the role of accounting valuation feedback.

In this paper, we triangulate hyper-inflation, inter-temporal shifts in accounting

standards, and the GFC to evaluate accounting valuation feedback on stock prices in Turkey.

Our research most closely aligns to Filip and Raffournier (2010) who undertook a seminal

study on the impact of inflation adjusted accounts on Bucharest Stock Exchange prices.

Turkey provides a natural experiment extension as a transition economy that experienced

annual hyper-inflation rates approaching 100% during the 1990’s, has high scores on the

Hofstede secrecy index, experienced several interventions from the International Monetary

Fund (IMF), and dramatic stock declines during the global financial crisis (GFC). Not

3

withstanding, Turkey experienced high annual growth rates of 5.3% (1997 to 2015), it is

situated in an economic and politically strategic position between Europe and Asia, it is

ranked by the World Bank as the 17th largest economy in the world, and 6th in Europe where it

is currently an EU candidate for full membership. Domestic (international) investors who

wish to invest (diversify) require knowledge of firm productive outcomes and the impact from

endogenous interruptions. Turkish authorities were aware that quality financial reporting

provides investors with a base to predict permanent earnings and fundamental value and high

price inflation did induce a re-evaluation of domestic Turkish financial reporting standards.

As a consequence, inflation accounting was in 2003 followed by IFRSs in 2005.1

Using quarterly data from 1997 to 2012 and, taking into account the post 2003

accounting interventions, consistent with western research we find significant positive

relationships between earnings, net book values and stock prices, with only an increase in

value relevance of net book values after 2003. In addition, inflation and real interest rates

provided valuation impact over and above accounting outputs. The window around the GFC

is a value relevance story for earnings. Before the GFC (2003-2007) earnings imposed a

strong positive signal on stock prices, but afterwards (2008-2012) the impact was essentially

reversed to zero.

Our paper makes several contributions. It is one of the few studies to clearly

highlight the influence of exogenous factors such as inflation, real interest rates, the GFC, and

intertemporal changes in accounting standards. Second, the GFC analysis reconciles the

positive coefficient results on earnings obtained by Türel (2009) (data ended in 2006), and the

negative results on earnings by Suadiye (2012) using a later data set. This result especially

1 Both accounting systems are conceptually argued—inflation accounting to reflect real value and productivity output (Edwards and Bell, 1961; Sweeney 1964), and IFRS ‘fair values’ to better reflect firm specific fundamentals (Dye and Sridhar, 2004; Barth et al., 2008; Ding et al., 2009).

4

highlights the importance of applying longer data sets with due account of exogenous

economic disruptions, and the role that accounting plays before and after. Finally, because of

differing culture, business conditions and a variation in the application of accounting

regulations, it is necessary to treat transition economies as individual case studies.

The paper now proceeds as follows. Section 2 provides an economic and accounting

background, section 3 develops research questions, and section 4 describes the data and

outlines the econometric tests. Section 5 contains the results and section 6 concludes the

paper.

2. ECONOMIC BACKGROUND AND ACCOUNTING INTERVENTIONS

2.1. The Turkish Economy

During the 1990’s the Turkish economy witnessed several economic upheavals including

falling GDP, hyper-inflation (CPI average 78.2%) that reached 120.3% in 1994. This was

quickly followed by a depreciation of the Turkish Lira by almost 70% against the US dollar,

and a Central Bank’s intervention in foreign exchange markets that caused a loss of more than

half of Turkey’s international reserves. This currency crisis was then followed by a banking

crisis where overnight interest rates jumped to over 50% and economic growth declined by

6% (Özatay, 2000). As a result, a stabilization program supported by an IMF stand-by

agreement was introduced in 1994.

Later, in 1998 and 1999, there were negative impacts from the Asian and Russian

financial crises and from two earthquakes that occurred in the Marmara region that impacted

almost half of the nation’s industrial output (Bibbee et al., 2010). This was followed by

negative growth rates, continued inflation rates over 60%, and a “disinflation programme”

5

accompanied by the 17th stand-by agreement introduced by the IMF at the end of 1999.2 To

cap it off, a liquidity shortfall emerged in November 2000 with a capital outflow crisis, a

currency devaluation of 50%, and a negative balance of balance of payments account. This

led to 19 banks (5 of which were listed banks), taking huge foreign exchange losses as a result

from the depreciated Turkish Lira. These banks were progressively taken over by the Savings

Deposit and Insurance Fund,3 and the crawling peg exchange rate system was replaced with a

floating exchange rate.

This litany of economic dilemmas, however, did not have a dampening effect on the

Istanbul Stock Exchange (IMKB) indicator stock index BIST100.4 Figure one shows that the

total market value increased from a low of 1441 to around 19526 over the 1997 to 2003

period (average 45% p.a.), despite high average annual inflation and interest rates of over

30%. Post 2003 also witnessed continued stock market gains over the 2004 to 2012 period

but with much greater price volatility at over three times the volatility of pre 2003 prices.5

This period also encompassed a final major economic event during our research period—the

impact of the GFC—which foresaw a dramatic decline in the IMKB. Effectively this

occurred over the period October 2007 to march 2009 when the IMKB lost 65% of its market

value and was associated with relatively low average inflation (8.6%) and average interest

rates of 13%. Figure 1 plots inflation rates, interest rates the IMKB index, and labels three

periods of interest (explained below) as pre accounting interventions (ACI) in the form of

inflation accounting and IFRS, with post ACI split into pre and post GFC. Of particular note

is the continued strength of the IMKB (average annual return 28.3%) and the dramatic decline

in inflation and interest rates.

2 Under this agreement, a crawling peg exchange rate system was applied, along with structural reforms, privatizations and a downsizing of the public sector. 3 Under control of the Banking Regulation and Supervision Agency (August 31, 2000) within the framework of Banks Act No. 4389. 4 Istanbul Stock Exchange has now changed its name to Borsa Istanbul A.S. 5 Average individual stock price volatility was 2.69 (see Table 2).

6

[INSERT FIGURE 1 ABOUT HERE]

2.2. Accounting Interventions (ACIs)

Pre 2003, Turkish companies prepared their financial statements in accordance with the

Turkish Uniform Accounting System based on historic cost and as a basis for tax calculations.

In addition, IMKB listed companies were required to prepare their financial statements in

accordance with the CMB’s Communiqué Serial: XI, No:1 “The Communiqué About the

Principals and Rules About Financial Statement and Reports in Capital Markets”. These

requirements created two different financial statements for IMKB companies—one prepared

basically for tax purposes and the other one prepared for capital markets purposes.

Induced by continued economic crises and high inflation, the relevance and reliability

of the financial accounting information disclosed over the pre-2003 period was perceived as

producing heavily distorted financial statements (Arsoy and Gucenme, 2009). Further, noting

that Turkey over the 1980/1990 period had the highest comparative inflation rates,

Whittington et al. (1997) illustrated the impact of inflation adjustments on listed firms. They

found decreases in mean net assets growth from 57.7% to 28.8%, sales growth from 56.2% to

3.2%, earnings before tax/net assets ratio from 32.4% to 12.3%, and earnings after tax/net

worth from 25.5% to 9.4%.

These shortcomings were seen to curtail investor confidence in the Turkish stock

market because of the lack of accounting transparency and value relevance. At the same time,

taxation adjustments were implemented that included LIFO, accelerated depreciation,

revaluations and indexation of assets. In order to overcome perceived shortcomings the CMB

enforced several new accounting regulations in the following years. First, consolidation and

inflation accounting became effective in the December 2003 financial statements. These

involved inflation gain (loss) adjustments to net monetary items passed through the income

7

statement and non-monetary adjustments to balance sheet items applying Wholesale Price

Indexes (Arsoy and Gucenme, 2009, appendix A). Second, in 2005 inflation accounting was

revoked and financial accounting reporting was harmonized with European Union regulations

and IFRS was partially introduced.6 IFRS adoption was gradual with some immediate

adoption of some revised translations, with one by one progressive translations of IFRSs that

proceeded up to 2008.7

2.3. Background Research

This paper fits within the longer term value relevance paradigm generally used to examine the

association between accounting data and stock market prices. Ball and Brown (1968), first

showed that earnings increases (decreases) are associated with positive (negative) abnormal

returns over 12 months prior to earnings announcement. With the advent of Ohlson (1995)

the clean surplus relation between price levels and book value and earnings was demonstrated

as ( tjtjttjtttj EARNBVP ,,,2,,1,0, εδδδ +++= ). Collins et al. (1997) applied this model to

U.S. data decomposed into three component equations: i) the incremental explanatory power

of book values, ii) the incremental explanatory power of earnings, and iii) the explanatory

power common to both earnings and book values. They reported an incremental decline in

the value relevance of “bottom line” earnings, countered by an increase in the value relevance

of book values (see also Francis and Schipper, 1999).

More related to this study is the application of association studies to investigate the

6 Arsoy and Gucenme (2009) view the application of inflation accounting that increases valuation reliability and comparability as consistent with a convergence towards IFRS. In this sense, the move towards IFRS was a natural progression in market based accounting. 7 CMB regulates the principles and procedures of preparation, presentation and announcement of financial statements prepared by the entities with the Communiqué No: XI-29, "Principles of Financial Reporting in Capital Markets" ("the Communiqué"). The Communiqué is effective for the annual periods starting from 1 January 2008 and supersedes the Communiqué No: XI–25, "The Accounting Standards in the Capital Markets". According to the Communiqué, entities shall prepare their financial statements in accordance with International Financial Reporting Standards ("IAS/IFRS") endorsed by the European Union. Until the differences of the IAS/IFRS as endorsed by the European Union from the ones issued by the International Accounting Standards Board ("IASB") are announced by Turkish Accounting Standards Board ("TASB"), IAS/IFRS issued by the IASB shall be applied.

8

introduction of IFRS in Europe. The prevailing assumptions being that IFRS represents an

improvement in accounting quality with greater comparability that, in turn, boosts investor

confidence and trading liquidity—resulting in a lower cost of capital and increased price

efficiency in stock markets. Results are somewhat mixed. Jermakowicz et al. (2007) found

that IFRS adoption by German companies significantly increases value relevance, and

Armstrong et al. (2010) report increased overall value relevance for European firms after

IFRS, as does Devalle et al. (2010). In contrast, Clarkson et al. (2009) found that earnings

and book value per share values measured according to IFRS in Europe, display no marginal

improvement compared to those measured with local GAAP.

Armstrong et al. (2010) also made the observation that IFRS had a lower impact in

European code law countries with weaker enforcement regimes. Focusing on Southern

Europe countries with code law accounting, Devalle et al. (2010) reported decreased value

relevance in Spain and Italy after IFRS, with Calleo et al. (2007) showing similar results for

Spain.8 More recently, Marra and Mazzola (2014) relate IFRS adoption to board

effectiveness in Italy, with higher board governance increasing the monitoring of IFRS quality

and constraining earnings management. IFRS quality reached its peak at the time of adoption

but they observed a declining trend thereafter. These papers caution that results on IFRS

value relevance are contentious without a consideration of contextual macro-economic and/or

firm business conditions (see also Liu et al., 2011).

Two such contextual impacts are inflation, which reduces the reliability of earnings

and net assets as valuation instruments (Board and Walker, 1990), and the level of nominal

interest rates that have a negative effect on market values independent of inflation and real

interest rates (Nissim and Penman, 2003). The papers on IFRS adoption in Europe ignore the

8 Readers are referred to Brüggemann et al. (2013) who provide a comprehensive review of the intended and unintended consequences of mandatory IFRS adoption in Europe.

9

impact of inflation on valuation reflecting the fact that inflation levels were generally low in

Western Europe over the last 25 years.9 This was not the case in transition and emerging

countries where inflation rates were high and approached hyper-inflation in some.

The paper by Filip and Raffournier (2010) is one of the few that makes direct

adjustments to account for the effects of hyper-inflation (average 48.7% p.a.) in Romania—

across a time period that experienced shifting measures of productivity signaled by several

dramatic changes in their accounting system. In this respect our paper is most closely aligned

to Filip and Raffournier (2010), who re-iterate the arguments that in emerging economies

accounting is the main source of productive information and research methodology must also

adjust for the macro-specifics of each market.

None of the prior papers on Turkish accounting take into account the impact of

inflation. For, example Türel (2009) applied the Ohlson model to examine the impact of

IFRS accounting on annual prices over the 2001-2006 period and reported, in aggregate, an

increased earnings coefficient with decreased value relevance of book values. Similarly,

Suadiye (2012) using the Ohlson model over 2000-2009 period, reported a reduction in

earnings and book coefficients after IFRS in 2005. Karǧın (2013) with an increased annual

data set from 1998-2011, reported increased value relevance of book values with no improvement

in the value relevance of earnings after IFRS. Bilgic and Ibis (2013) over a similar period 1997-

2011, report an increase in the value relevance of book values with a decrease in value relevance

of earnings after IFRS. Other research takes different perspectives. Gökmen (2013) examined

the value relevance of earnings after IFRS using conditional conservatism and found reduced

value relevance in manufacturing firms, and Dinçer (2009) reported a reduction in the adverse

selection cost component of the bid-ask spread after IFRS adoption.

9 For example, annual inflation between 2-3% for Germany, France and the U.K. over the last 25 years.

10

In summary, the Turkish results are conflicting, probably caused by several factors

that we subsequently take into account. First, no adjustments for inflation and (real) interest

rates were imposed. The average annualized inflation was 60.9% over part of the research

periods (1997-2003) with average interest rates of 56.8%. Inflation dropped thereafter but

still maintaining annual rates approaching 9% associated with relatively higher interest rates

of 13.1% (see Table 2). Second, the several changes in the macro-economic environment and

accounting reporting which included the GFC impact in 2007, was ignored. Finally, annual

data was used, but Turkish accounts are available on a quarterly basis and this provides a

more timely assessment of value relevance and conditional inflationary impact.

3. RESEARCH QUESTIONS

Our first research questions follow prior research in addressing whether the market based ACI

accounting standard changes in 2003 and 2005 increased value relevance. They include CMB

accounting adjustments for general inflation and direct fair value adjustments to assets,

liabilities and earnings under IFRS. Both purport to add value relevant information for

investors and, conceptually, if these accounting adjustments represent unobserved economic

phenomena they, are ‘truer’ estimates of fundamental value.

However, we do not see the accounting/price relationship as that simple. A broad

claim that inflation and price adjustments provide value relevant information to a wide set of

users may be more hopeful rather than informative. Fundamental accounting feedback

depends not only on the quality of accounting information capture, but on firm specific

characteristics. For example, given the Turkish situation that consists of a high level of

family based ownership and internalized information flow, one might expect price setting to

rely more on industry and market wide sources. This induces two competing scenarios—

either there will be a higher level of stock synchronicity with investors apprehensive about

inflation (fair value) constructed accounts, or enhanced firm accounting quality from enforced

11

upgraded standards. Given, the uncertainty about direction and the conflicting past research

(Türel, 2009; Suadiye, 2012; Karǧın, 2013, Bilgic and Ibis, 2013) we propose two non-

directional research questions:

RQ1: Was there a positive value relevance association between earnings, book values and stock prices in Turkey, and did the introduction of inflation accounting and IFRS increase these associations?

RQ2: What accounting component(s) provided the highest valuation impact?

Once these common research questions are imposed, we are further interested in

asking whether there is incremental information contained in the level of inflation and the real

rate of interest. For example, if inflation is fully and specifically incorporated into accounting

numbers then the general level of inflation should contain no further “value based”

information. We are particularly interested in two aspects of the Turkish inflation level—the

impact from the high level of inflation over the earlier part of our data pre 2003, and the

association between inflation and interest rates.

RQ3: Did inflation play a residual role before and after the accounting adjustments?

Moreover, developing markets have a higher level of risky industries, higher

inflation, and are distinguished by lower levels of financial development and volatile markets.

Given these attributes, Ball (2008) argues that a greater reliance on fair value adjusted

accounting is inappropriate for countries with poorly functioning financial and capital

markets. Taking this a step further, if Turkey has a secretive culture with lower levels of

accounting enforcement then there is a possibility of accounting management that inflates

earnings and book values that provides in-appropriate signals from the now revised and

authoritative published accounting reports. In this regard, Laux and Leuz (2009) suggest that

fair value adjustments contain contagion effects which cause overreaction of market

participants, in turn, reducing both the firm specific and market wide reliability of financial

12

accounting (see also Plantin et al., 2008).10 The GFC recession had a major impact on the

Turkish stock market with a decline in value of about two-thirds during 2008. We propose to

examine this issue by decomposing the data into the period before the stock market decline

and afterwards, by posing the following two research questions:

RQ4: What was the feedback association from accounting before the GFC period?

RQ5: What was the feedback association from accounting during the GFC period?

4. DATA AND MODELS

4.1. Data

In order to make the results comparable throughout the years, a balanced panel of surviving

IMKB companies over the period 1997-2012 was selected. From these companies, only those

with final December year end accounting periods were retained and companies from regulated

sectors, including banking, insurance, leasing, and factoring were excluded. Also excluded

were companies placed on the IMKB Watchlist and/or companies whose share trading was

temporarily suspended by the IMKB during financial statement announcement dates.11

Finally, for the quarterly data for each variable to be tested, outlier observations outside the

range of plus or minus three standard deviations from the mean were excluded. The final

sample consisted of 113 companies with industrial distributions reported in Table 1.

[INSERT TABLE 1 ABOUT HERE]

In 2005, the unit of exchange was changed to New Turkish Lira (NTL), which is

equivalent to 10-6 times (old) Turkish Lira. Hence, all data for the effected periods were

divided by 106 to make them comparable across periods, with prices adjusted to exclude the

10 This is a subtle argument. If increased prices result from demand induced by productive output then they reflect fundamental economic factors. But if they reflect over-exuberance then the price signal is vacuous at best, or irrational noise at worse and if receiving the stamp of international approval then they become more authentic and believable. 11 As a result of accounting manipulation investigations and corporate financial problems.

13

effects of stock splits, dividend payments, and share and bonus issues. All variables are then

normalized by dividing through by the average number of shares outstanding during the

related quarter. We also hand collect the actual date of public release of the financial

statements and match with the next day’s closing price to allow for a potential lag in the stock

price adjustment.12 Finally for testing, data is divided into a number of sub-periods: 1997Q1

to 2003Q3 (pre-accounting intervention); 2003Q4 to 2012Q4 (inflation and IFRS

intervention); 2003Q4 to 2007Q4 (pre GFC stock decline); 2008Q1 to 2012Q4 (post GFC

decline).

Descriptive statistics across the pre and post accounting interventions are contained

in Table 2. Comparing the post accounting period the key features are a significant increase

in stock market volatility, reduced earnings and total liabilities with increased net book values

and lower inflation and interest rates. Correlation coefficients are contained in Table 3. The

single notable change over the two periods is the switch from positive to negative for the

correlation between inflation and interest rates to prices.

[INSERT TABLES 2 & 3 ABOUT HERE]

4.2. Statistical Models

We build and analyze five different models stock price valuation models using firm fixed

effect regressions. The first model is a simple application Ohlson’s model (1995) that

analyses earnings and net book impact on prices with interactive terms to test the difference of

value price relevance after the introduction of inflation and IFRS accounting. Model (2)

decomposes net book values into total assets and total liabilities to test separate explanatory

power. In Model (3), we add the inflation rate to assess incremental impact not captured by

12 Up until 2009 Q2, financial statements were announced to the public by means of IMKB Daily Bulletins after the second trading session ended and then disclosed on the IMKB website. Since the financial statements are effectively disclosed to the public with a one day lag we adjust the price impact to take this into account.

14

any accounting adjustments, and Models (4) and (5) examine the incremental interaction with

the real cost of debt.

Pj,t+1 = α0 + α1 EARNj,t + α2 EARNj,t × Dj,t + α3 BVj,t + α4 BVj,t × Dj,t + Fixed Effects + εj,t (1)

Pj,t+1 = β0 + β1 EARNj,t + β2 EARNj,t × Dj,t +β3TAj,t + β4TAj,t × Dj,t

+ β5TLj,t + β6 TLj,t × Dj,t+ Fixed Effects + εj,t (2)

Pj,t+1 = θ0 + θ1 EARNj,t + θ2 EARNj,t × Dj,t + θ3 TAj,t + θ4 TAj,t × Dj,t

+ θ5 TLj,t + θ6 TLj,t × Dj,t+ θ7 INFLj,t + θ8 INFLj,t × Dj,t

+ Fixed Effects + εj,t (3)

Pj,t+1 = γ0 + γ1 EARNj,t + γ2 EARNj,t × Dj,t + γ3TAj,t + γ4TAj,t × Dj,t + γ5 TLj,t + γ6 TLj,t × Dj,t + γ7 DIFF_Pj,t + γ8 DIFF_Pj,t × Dj,t

+ Fixed Effects + εj,t (4)

Pj,t+1 = τ0 + τ1 EARNj,t + τ2 EARNj,t × Dj,t + τ3TAj,t + τ4 TAj,t × Dj,t + τ5 TLj,t + τ6 TLj,t × Dj,t + τ7 DIFF_Nj,t + τ8 DIFF_Nj,t × Dj,t

+ Fixed Effects + εj,t (5)

where:

Pj,t+1: Closing share price of company j at the financial statement announcement day at quarter t . EARNj,t: Net earnings after tax of company j at the financial statement announcement day at quarter t. BVj,t: Net book value of company j at the financial statement announcement day at quarter t. TAj,t: Total assets of company j at the financial statement announcement day at quarter t. TLj,t: Total liabilities of company j at the financial statement announcement day at quarter t. INFLj,t: Average inflation rate expressed on a yearly basis at quarter t. DIFF_P: A dummy variable equal to 1 when INFL minus the Turkish Central Bank benchmark quarterly deposit interest rate is positive. Otherwise, 0. DIFF_N: A dummy variable equal to 1 when INFL minus the Turkish Central Bank benchmark quarterly deposit interest rate is negative. Otherwise, 0.

15

Dj,t: A dummy variable related to either the entire post inflation/IFRS time period (>2003Q4), or the GFC period (>2007Q4).

Firm fixed effects regressions are run for several substantial reasons. The first is to

control for selection bias. When running regressions, if there are some omitted firm

characteristics that drive both net book values and earnings, the results could be entirely

explained by a biased firm selection. Similarly, some firms may have a positive relevance

between book value and earnings and stock prices, while some may exhibit no (negative)

relationship. Finally, firms often possess idiosyncratic features and business operations. By

including firm fixed effects, we control for time-invariant differences across firms.

5. RESULTS

5.1. Pre and Post Accounting Interventions

The first focus is on RQ1 and RQ2 as reflected in models (1) and (2) as detailed in Table 4.

Model 1 shows an overall positive relation between earnings and stock prices and net book

value and stock prices. When we then apply an interactive dummy variable to account for the

ACI introductions of inflation and IFRS accounting from December 2003, there is a

significant increase in the net book value coefficient (0.775, t=3.52) but no significant

increase in the explanatory power of earnings (1.31, t=0.55). This result confirms Karǧın

(2013) who used a longer data set and Bilgic and Ibis (2013), but is inconsistent with Türel

(2009) and Suadiye (2012). Hence, ACI adjustments to book values contain the highest

information content.

To extract the components that have the greatest impact, we decompose net book

value into assets and liabilities (Model 2), and see that both have the expected signs (+ assets,

- liabilities). After the adjustments imposed by ACI both dummy coefficients are significant,

but with a higher negative impact for liabilities. At face value, these results are consistent

with the following: (i) as argued by the IASB, adjustments to assets that reflect current

16

opportunity cost/market values contains increased price relevant information to investors, and

(ii) holding liabilities in lower inflationary periods is more costly especially if real interest

rates are relatively higher.

[INSERT TABLE 4 ABOUT HERE]

Our next equation (model 3) addresses the question as to whether general inflation

had any further incremental valuation impact over and above the inflation and price

adjustments directly incorporated by the ACIs. We find a positive impact of inflation on

prices before 2003 (0.101, t=5.44), but a significant reduction afterwards (-0.357, t=2.62).

There are several explanations. First, average annualized inflation before 2003 was 60.9%

compared to 8.6% afterwards. High inflation reduces the real value of nominal debt holdings,

encourages the expansion of debt with repayment at lower real values. We explore this

further by examining total firm liability holdings and observe they were reduced after 2003,

from 7.30 to 5.74 million Turkish lira (Table 2), representing a 21.4% fall in average debt.

The above explanations are conditional on the real cost of debt. If real interest rates

are negative they will induce a higher demand for debt and vice versa if real interest rates are

positive. In equation (4) we consider what happens when inflation rates are higher than

interest rates by examining the DIFF_P variable. There is a strong positive impact from

DIFF_P before 2003 (8.62, t=10.20), with a significant reduced impact afterwards (-3.98,

t=2.68), consistent with a prolonged period of negative real interests that induce higher

demand for debt’ which is then associated with increased firm value. Model (5) examines the

case when interest rates are higher than inflation which potentially negates valuation returns

from inflation. We report a negative valuation impact (-4.29, t=5.08), which increases after

2003 when real interest rates exhibited a more consistent positive trend. In short, in terms of

RQ3, accounting adjustments alone do not serve to capture the full effects of inflation and real

17

interest rates. Taken together, markets are aware of the valuation impacts of inflation and

incrementally incorporate these into prices both before and after accounting interventions that

make specific market based adjustments.

One aspect of our analysis is that the post 2003 period lumped together the

adjustments incorporating both inflation accounting and IFRS. There may be differential

impacts from general inflation adjustments compared to firm specific fair value adjustments

under IFRS. We check this by rerunning all models in Table 4 with an additional interactive

dummy from the introduction of IFRS in 2005. There was no significant difference in the

value relevance impact coefficients for earnings, comprehensive earnings adjustments, net

book values, inflation, or real interest rates between the two different accounting models.

5.2. Pre and Post GFC

Our further analysis addresses RQ4 and RQ5 by taking into account the GFC which saw the

bursting of the asset bubble and a loss of 65% of Turkish stock market value.13 To undertake

this analysis we compress the data into the period from the introduction of inflation

accounting in 2003Q4 and separate before and after the decline in the stock market from

2007Q4.

[INSERT TABLE 5 ABOUT HERE]

Table 5 reports the results. The basic model (1) shows a significant increase in net

book values after advent of the GFC as does the separate asset component in model (2).

Liabilities show no statistically significant change. The more stark and interesting result is

the dramatic shift in the earnings’ coefficient. They are large and strongly positive before the

GFC (e.g. Model (1) 10.55, t=3.98) with a sharp reverse afterwards (-11.79, t=4.34). This

result demonstrates and explains the previous positive results on earnings obtained by Türel

13 This compares with a loss of 47% over a similar period for the FTSE100.

18

(2009) using a data set that ended in 2006 and reconciles the negative results on earnings by

Suadiye (2012). Models (3, 4 & 5) incorporate inflation and real interest rates to show no

significant impact before the GFC, but significance afterwards. This impact was, in the main,

driven by higher levels of real interest rates (-7.01, t=5.08), signifying that the market was

now very price wary of firms that held debt that attracted real rates of interest.

5.3. Direct Adjustments for Inflation

Our model differs from the analysis of Filip and Raffournier (2010) who apply a change

model and make direct adjustments for inflation to calculate adjusted returns, adjusted

earnings and adjusted change in earnings (see Eq. 2, p.92). In order to test the robustness of

our results we replicate their adjustments and rerun the Turkish data over the same time splits

using OLS and panel regressions. The OLS coefficients of about 0.4 for adjusted earnings

and 0.01 for earnings changes for the pre 2003 period are comparatively similar to Filip and

Raffournier. After splitting the data into the pre (post) GFC periods we obtained the same

reversal impact, with adjusted earnings coefficients of 0.31 (-0.29) and adjusted earnings

change coefficients of 0.02 (-0.02). Very similar results were obtained from the fixed effect

regressions.

6. CONCLUSION AND DISCUSSION

This paper evaluates the stock valuation impact of accounting data in Turkey across three

significant economic events—high inflation, the real cost of debt and the GFC. These events

are overlaid with the accounting interventions by the Turkish authorities that changed the

historic cost accounts by adjustments for inflation (2003) and the introducing IFRS (2005).

Using quarterly data from 1997 to 2012 we find significant positive relationships

between earnings and net book values with stock prices, with only an increase in value

relevance of net book values after 2003. This aggregate result is similar to Francis and

19

Schipper (1999) who detected a rising trend in the value relevance of net book values over

time, which also reinforces the IASB’s push towards a more balance sheet valuation focus for

accounts. In addition, before 2003, inflation had a positive impact on value, over and above

accounting outputs (driven by negative real interest rates), and a negative value impact after

2003 (driven by positive real interest rates). This reveals the changing impact from inflation

and interest rates and cautions that accounting coefficients are potentially mis-specified

without the consideration of potential macro impacts.

After narrowing the test window around the GFC, net assets continue to increase in

value relevance after 2007; in conjunction with a negative impact from positive real interest

rates that signals a valuation penalty for firms holding higher debt. What is more interesting

is the divergent value relevance of earnings. Before the GFC (2003-2007) earnings sent a

strong positive valuation signal to stock prices, but afterwards (2008-2012) the impact was

essentially zero. This part of the study reconciles the positive coefficient results on earnings

obtained by Türel (2009) using a data set that ended in 2006, and the negative results on

earnings by Suadiye (2012) using a later data period. A reinforcement of the danger of

producing conflicting results from dissimilar inter-temporal data sets.

A reflective question is to ask whether the introduction of inflation and fair value

adjusted accounting improved the information environment for Turkish shareholders?

Regarding the balance sheet there was continued and stable increase in valuation feedback.

However, the conflicting nature of earnings around the GFC raises concerns about whether

accounting, in an overall sense, improved valuation fundamentals? The shifting nature of the

earnings coefficient, by itself, indicates that the updated Turkish accounting systems was not a

leading indicator. At an extreme one might suggest pre-GFC earnings were associated with

over-optimistic perceptions of value. At a minimum, questions are raised as to whether

accounting plays a first-order valuation role in countries with below average firm and country

20

wide governance structures. These issues provide further outlets for research endeavor.

21

REFERENCES

Armstrong, C. S., Barth, M. E., Jagolinzer, A. D., & Riedl, E. J. (2010). Market reaction to the adoption of IFRS in Europe. The Accounting Review, 85(1), 31-61.

Arsoy, A. P., & Gucenme, U. (2009). The development of inflation accounting in Turkey. Critical Perspectives on Accounting, 20(5), 568-590.

Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers. Journal of Accounting Research, 6(2), 159-178.

Ball, R. (2008). “Don’t Blame the Messenger … or Ignore the Message”. http://www.chicagobooth.edu/email/chicagoon/ShootingTheMessenger10.12.2008.pdf

Barth, M. E., Landsman, W. R., & Lang, M. H. (2008). International accounting standards and accounting quality. Journal of Accounting Research, 46(3), 467-498.

Bibbee, A., Gonenc, R., Jacobs, s., Konvits, J., & Price, R. (2000). Economic effects of the 1999 Turkish earthquakes: An interim report. http://www.spo.org.tr/resimler/ekler/fecdb276f634854_ek.pdf?tipi=58&turu=X&sube=0

Bilgic, F. A., Ibis, C. (2013). Effects of New Financial Reporting Standards on Value Relevance, A Study About Turkish Stock Markets. International Journal of Economics and Finance, 5(10), 126-140. Board, J., & Walker, M. (1990). Intertemporal and cross-sectional variation in the association between unexpected accounting rates of return and abnormal returns. Journal of Accounting Research, 28, 182-192. Brüggemann, U., Hitz, J.-M., & Sellhorn, T. (2013). Intended and unintended consequences of mandatory IFRS adoption: A review of extant evidence and suggestions for future research. European Accounting Review, 22(1), 1-37. Clarkson, P., Hanna, J. D., Richardson, G. D., & Thompson, R. (2011). The impact of IFRS adoption on the value relevance of book value and earnings. Journal of Contemporary Accounting and Economics, 7(1), 1-17. Collins, D. W., Maydew, E. L., & Weiss, I. S. (1997). Changes in the value-relevance of earnings and book values over the past forty years. Journal of Accounting and Economics, 24(1), 39-67. Devalle, A., Onali, E., & Magarini, R. (2010). Assessing the value relevance of accounting data after the introduction of IFRS in Europe. Journal of International Financial Management and Accounting, 21(2), 85-119.

Dinçer, B. (2009). The impact of adoption of IAS/IFRS on information asymmetry in Istanbul Stock Exchange, YÖK.

Ding, Y., Jeanjean, T., & Stolowy, H. (2009). Observations on measuring the differences between domestic accounting standards and IAS: a reply. Journal of Accounting and Public Policy, 28(2), 154-161.

22

Dye, R. A., & Sridhar, S. (2004). Reliability-Relevance Trade-Offs and the Efficiency of Aggregation. Journal of Accounting Research, 42(1), 51-88.

Edwards, E. O., & Bell, P. W. (1961). The theory and measurement of business income. Berkeley [Calif.] Book: University of California Press. Filip, A., & Raffournier, B. (2010). The value relevance of earnings in a transition economy: the case of Romania. The International Journal of Accounting, 45, 77-103. Francis, J., & Schipper, K. (1999). Have financial statements lost their relevance? Journal of Accounting Research, 37(2), 319-352.

Gökmen, M. K. (2013). An empirical study on conditional conservatism and value relevance of earnings. International Journal of Social Science, 6(2), 1697-1720.

Jermakowicz, E. K., Prather-Kinsey, J., & Wulf, I. (2007). The value relevance of accounting income reported by DAX-30 German companies. Journal of International Financial Management and Accounting, 18(3), 151-191. Karǧın, S. (2013). The impact of IFRS on the value relevance of accounting information: Evidence from Turkish firms. International Journal of Economics and Finance, 5(4), 71-80. Laux, C., & Leuz, C. (2009). The crisis of fair-value accounting: making sense of the recent debate. Accounting, Organizations and Society, 34(6/7), 826-834. Marra II, A., & Mazzola, P. (2014). Is corporate board more effective under IFRS or ‘Its Just an Illusion’? Journal of Accounting, Auditing and Finance, 29, 1, 31-60. Nissim, D., & Penman, S. (2003). Associations between changes in interest rates, earnings, and equity values. Contemporary Accounting Research, 20, 775−804. Ohlson, J. A. (1995). Earnings, book values, and dividends in equity valuation. Contemporary Accounting Research, 11(2), 661-687.

Özatay, F. (2000). The 1994 currency crisis in Turkey. Journal of Policy Reform, 3(4), 327-352.

Plantin, G., Sapra, H., & Shin, H. S. (2008). Marking-to-market: Panacea or pandora's box? Journal of Accounting Research, 46(2), 435-460. Suadiye, G. (2012). Value relevance of book value and earnings under the local GAAP and IFRS: Evidence from Turkey. Ege Academic Review, 12, 301-310. Sweeney, H.W. (1964). Stabilized Accounting, New York: Holt, Rinehart and Winston Türel, A. (2009). The value relevance of IFRS: The case of Turkey. Acta Universitatis Danubius: Oeconomica, 5, 119-128. Whittington,G., Saporta,V., & Singh, A. (1997). The effects of hyper-inflation on accounting ratios: Financing corporate growth in industrial economies.

23

http://wwwwds.worldbank.org/external/default/WDSContentServer/WDSP/IB/1997/08/01/000009265_3980313102034/Rendered/PDF/multi0page.pdf

24

Figure 1: IMBK BIST100 Index, Inflation and Interest Rates

0

20

40

60

80

100

120

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

Infla

tion/

Inte

rest

Rat

es

BIS

T 1

00 In

dex

Val

ue

Date

BIST 100 INTEREST(%) INFLATION(%)

ACI Prior to GFC Prior to ACI

GFC and Post GFC

25

Table 1: Industrial Distribution of Sample Companies

Sectors Number of Companies % Non-Metallic Mineral Products 19 16,81 Chemicals, Petroleum, Rubber and Plastic Products 16 14,16 Fabricated Metal Products, Machinery and Equipment 15 13,27 Food, Beverage and Tobacco 12 10,62 Textile, Wearing Apparel and Leather 12 10,62 Basic Metal Industries 10 8,85 Wood, Paper and Printing 9 7,96 Services 8 7,08 Holding and Investment Companies 6 5,31 Technology 3 2,65 Construction and Public Works 2 1,77 Real Estate Activities 1 0,88 Total 113 100

Source: www.borsaistanbul.com.trr

26

Table 2: Descriptive Statistics This table reports the descriptive statistics for the quarterly sample data over the period from 1997-2012. PRICE is the stock price one day after the earnings announcement in new Turkish Lira. EARN is quarterly earnings, TA is quarterly total assets, and TL is quarterly total liabilities, all expressed in new Turkish Lira (millions) and standardized by quarterly average number of shares outstanding. INFL is quarterly annualized inflation rate, and INTE is the Turkish Central Bank benchmark deposit interest rate. BIST100 is the closing quarterly market stock price index.

Panel A: Before Accounting Initiatives (1997Q1 – 2003Q3)

Mean Std. Dev Min Max PRICE 14.75 23.03 0.35 305 BIST100 8762.58 4774.034 1441.19 18470.06 EARN 0.71 2.08 -9.19 29.63 BV 5.49 7.07 -4.52 78.17 TA 12.96 12.43 0.47 102.52 TL 7.30 7.35 0.00 60.30 INFL (%) 60.94 21.43 21.12 99.35 INTE (%) 56.86 15.37 25.73 78.67

Panel B: After Accounting Initiatives (2003Q4 – 2012Q4)

Mean Std. Dev Min Max PRICE 16.00 62.14 0.22 1290 BIST100 44713.44 16145.11 17786.42 78166.25 EARN 0.47 1.90 -43.96 32.86 BV 7.56 13.06 -2.23 113.02 TA 13.29 18.52 0.00 154.92 TL 5.74 8.09 0.00 97.55 INFL (%) 8.58 1.92 4.35 14.24 INTE (%) 13.11 6.05 5.75 27

27

Table 3: Correlation Matrixes

This table reports the correlation for the quarterly sample data over the period from 1997-2012. PRICE is the stock price one day after the earnings announcement in new Turkish Lira. EARN is quarterly earnings, TA is quarterly total assets, and TL is quarterly total liabilities, all in new Turkish Lira (Million) and standardized by quarterly average number of shares outstanding. INFL is quarterly annualized inflation rate, and INTE is the Turkish Central Bank benchmark deposit interest rate. BIST100 is quarterly market stock price index. * p<0.1, ** p<0.05, *** p<0.01 indicate significance levels.

Panel A: Before Accounting Initiatives (1997Q1 – 2003Q3)

PRICE BIST100 EARN BV TA TL INFL INTE PRICE 1.0000*** BIST100 0.0195*** 1.0000*** EARN 0.5312*** 0.0716*** 1.0000*** BV 0.5323*** -0.0530*** 0.3999*** 1.0000*** TA 0.5829*** -0.0159*** 0.4203*** 0.8436*** 1.0000*** TL 0.4552*** -0.0219*** 0.2383*** 0.4824*** 0.8702*** 1.0000*** INFL 0.1519*** -0.6036*** 0.0946*** -0.0218*** 0.0201*** 0.0495*** 1.0000*** INTE 0.0022*** -0.7003*** 0.0839*** -0.0244*** 0.0278*** 0.0550*** 0.6012*** 1.0000***

Panel B: After Accounting Initiatives (2003Q4 – 2012Q4)

PRICE BIST100 EARN BV TA TL INFL INTE PRICE 1.0000*** BIST100 0.0485*** 1.0000*** EARN 0.5131*** 0.0221*** 1.0000*** BV 0.6515*** -0.0138*** 0.6081*** 1.0000*** TA 0.5592*** 0.0046*** 0.5523*** 0.9265*** 1.0000*** TL 0.2279*** 0.0292*** 0.2838*** 0.5097*** 0.7950*** 1.0000*** INFL -0.0203*** -0.4160*** 0.0175*** 0.0036*** 0.0037*** 0.0048*** 1.0000*** INTE -0.0333*** -0.8492*** 0.0098*** 0.0227*** -0.0005*** -0.0333*** 0.4907*** 1.0000***

28

Table 4: Before and After the ACIs: Balanced Panel Regression Analysis (Period: 1997Q1 – 2012Q4) This table reports firm fixed effects regression results using quarterly stock prices as dependent variables for the period from 1997Q1 to 2012Q4. EARN is quarterly earnings, BV is net book value, TA is quarterly total assets, and TL is quarterly total liabilities, all in new Turkish Lira (Million) and standardized by quarterly average number of shares outstanding. INFL is quarterly inflation rate, and DIFF_P (DIFF_N) is when INFL minus the Turkish Central Bank benchmark quarterly deposit interest rate is positive (negative). ACI is a dummy variable equal to 1 for the period 2003Q4 – 2012Q4 indicating the period of inflation and IFRS accounting. EARN×ACI, CEARN×ACI, BV×ACI, TA×ACI, TL×ACI, INFL×ACI and DIFF_P×ACI are interaction terms. t-statistics are in parentheses and * p<0.1, ** p<0.05, *** p<0.01 indicate significance levels. (1) (2) (3) (4) (5) EARN 3.474*** 2.908*** 2.850*** 2.978*** 2.973*** (5.90) (4.94) (4.95) (5.19) (5.14) EARN×ACI 1.131 1.335 1.218 1.154 1.097 (0.55) (0.71) (0.65) (0.61) (0.58) BV 1.962*** (11.13) BV×ACI 0.775*** (3.52) TA 1.359*** 1.273*** 1.332*** 1.310*** (6.58) (6.44) (6.64) (6.53) TA×ACI 1.999*** 2.163*** 2.071*** 2.114*** (6.32) (6.59) (6.51) (6.54) TL -0.751** -1.012*** -0.859*** -0.872*** (-2.35) (-3.10) (-2.71) (-2.73) TL×ACI -4.057*** -3.751*** -3.977*** -3.988*** (-7.19) (-6.92) (-7.20) (-7.15) INFL 0.101*** (5.44) INFL×ACI -0.357*** (-2.62) DIFF_P 8.619*** (10.20) DIFF_P×ACI -3.978*** (-2.68) DIFF_N -4.289*** (-5.08) DIFF_N×ACI -4.128*** (-3.48) Intercept -3.821*** -3.887*** -4.898*** -6.284*** 0.726 (-5.82) (-5.50) (-4.04) (-8.39) (0.82) Fixed Effects Firm Firm Firm Firm Firm N 7232 7232 7232 7232 7232 adj. R2 0.549 0.573 0.577 0.577 0.578

29

Table 5: Before and After the GFC: Balanced Panel Regression Analysis (Period: 2003Q4-2012Q4) This table reports firm fixed effects regression results using quarterly stock prices as dependent variables for the period from 2003Q4 to 2012Q4. EARN is quarterly earnings, BV is net book value, TA is quarterly total assets, and TL is quarterly total liabilities, all in new Turkish Lira (Million) and standardized by quarterly average number of shares outstanding. INFL is quarterly inflation rate, and DIFF_P (DIFF_N) is when INFL minus the Turkish Central Bank benchmark quarterly deposit interest rate is positive (negative). GFC is a dummy variable equal to 1 for the period 2007Q4 – 2012Q4 the period covering the GFC stock decline. EARN×GFC, CEARN×GFC, BV×GFC, TA×GFC, TL×GFC, INFL×GFC and DIFF_P×GFC are interaction terms. t-statistics are in parentheses and * p<0.1, ** p<0.05, *** p<0.01 indicate significance levels. (1) (2) (3) (4) (5) EARN 10.55*** 9.701*** 9.769*** 9.685*** 9.756*** (3.98) (3.95) (3.97) (3.94) (3.97) EARN×GFC -11.79*** -11.02*** -11.20*** -11.01*** -11.27*** (-4.34) (-4.35) (-4.42) (-4.35) (-4.44) BV 3.015*** (6.98) BV×GFC 1.421*** (4.67) TA 3.829*** 3.776*** 3.843*** 3.778*** (7.35) (7.26) (7.38) (7.27) TA×GFC 1.399*** 1.478*** 1.377*** 1.488*** (3.18) (3.29) (3.10) (3.30) TL -5.922*** -5.982*** -5.903*** -6.004*** (-6.45) (-6.47) (-6.43) (-6.48) TL×GFC -1.030 -0.887 -1.078 -0.901 (-1.27) (-1.11) (-1.34) (-1.13) INFL -0.184 (-0.64) INFL×GFC -0.556*** (-3.99) DIFF_P -4.223*** (-3.01) DIFF_P×GFC 2.796** (2.11) DIFF_N 1.517 (0.94) DIFF_N×GFC -7.076*** (-5.08) Intercept -6.449*** -7.679*** -3.495 -8.437*** -6.644*** (-6.63) (-7.04) (-1.31) (-8.09) (-3.74) Fixed Effects Firm Firm Firm Firm Firm N 4181 4181 4181 4181 4181 adj. R2 0.669 0.687 0.688 0.687 0.689